Introduction
Ulcerative colitis (UC), a chronic, relapsing-remitting inflammatory bowel disease (IBD), primarily affects the colonic mucosa and arises from a complex interplay of genetic predisposition, environmental triggers and immune dysregulation. Key pathogenic mechanisms include epithelial barrier dysfunction, aberrant immune responses and gut microbiota dysbiosis, which collectively perpetuate chronic inflammation.1 UC is associated with substantial morbidity, significantly impairs quality of life and incurs considerable healthcare costs.2 The therapeutic management of UC has shifted to treat-to-target strategies,3 moving beyond symptom-based care to objective, goal-oriented approaches. The Selecting Therapeutic Targets in Inflammatory Bowel Disease (STRIDE-II)4 recommendations outlined short-term targets (symptomatic relief), intermediate-term targets (clinical remission and biomarker normalisation) and long-term goals (endoscopic healing and restored quality of life) to achieve sustained disease control.
This strategic framework has spurred advances in targeted therapies, expanding options for precise disease management. The armamentarium for UC has now expanded to include biologics such as vedolizumab (VDZ), a gut-selective humanised monoclonal antibody that has demonstrated robust efficacy by selectively targeting the α4β7 integrin, thereby inhibiting lymphocyte homing to inflamed gastrointestinal mucosa.5 It has been shown to be effective as a first- or second-line treatment for both inducing and maintaining remission in UC,6 with a favourable safety profile supporting its long-term use. Despite its effectiveness, both primary non-response and secondary loss of response to VDZ occur.7 Real-world experience with VDZ in UC was recently reported in a systematic review, with clinical response and remission rates at week 52 (W52) of 48% and 39%, respectively.8 Therapeutic drug monitoring (TDM) is a cornerstone in IBD, with strong evidence supporting its use during induction and maintenance phases, as well as in proactive and reactive strategies.9 However, while well established for anti-tumour necrosis factor (anti-TNF) therapies, particularly infliximab, the evidence for the use of TDM for VDZ is comparatively less robust.7
Anti-integrin αvβ6 autoantibodies have recently emerged as a promising non-invasive biomarker class for diagnosing UC, potentially outperforming conventional biomarkers due to their high sensitivity and specificity.10 Their robust performance has been validated across different cohorts worldwide, including Japanese,11 Swedish,12 North American13 and Italian14 populations. Studies from multiple cohorts have shown that this serological biomarker is detectable up to a decade before the clinical onset of UC.13 The presence of αvβ6 autoantibodies has been shown to reflect both clinical and endoscopic disease activity,1516 and these autoantibodies are associated with adverse UC-related outcomes,13 positioning them as valuable tools for early detection and longitudinal disease monitoring.
The primary objective of this study was to validate the diagnostic accuracy of serum anti-integrin αvβ6 autoantibodies in a French multicentre cohort of patients with IBD. The secondary objectives were to assess whether baseline serum αvβ6 autoantibody titres and early changes in these titres could serve as predictive biomarkers of response to VDZ in patients with UC, to evaluate their association with VDZ pharmacokinetics and therapeutic efficacy and to determine their ability to predict sustained clinical remission at W52.
Materials and methods
Study design and clinical cohorts
The present study was a multicentre, non-interventional post hoc analysis of two prospective observational cohorts: VEDOPREDIRESP (patients with IBD initiating vedolizumab) and GOLILOR (patients with UC undergoing golimumab dose intensification). The present analysis used prospectively collected serum samples and clinical data from these cohorts to investigate the association between anti-αvβ6 immunoglobulin G (IgG) antibody dynamics and subsequent clinical outcomes.
The VEDOPREDIRESP parent study is a phase IV, multicentre, prospective observational study assessing the exposure–response relationship of intravenous VDZ (300 mg at W0, W2, W6, then every 8 weeks) in adults (≥18 years) with IBD and evaluating the relationship between induction serum trough concentrations and treatment outcomes over 52 weeks. Consequently, all patients underwent prospective blood tests at the specified time points, with serum samples collected for all patients. We performed a post hoc analysis using prospectively collected serum, faecal calprotectin and clinical data from the VEDOPREDIRESP cohort.
Patients were enrolled between 2016 and 2022 across five French tertiary centres: Saint-Étienne University Hospital, Lyon Sud University Hospital, Bicêtre University Hospital, Amiens University Hospital and Nice University Hospital. Eligible participants had moderate-to-severe UC (partial Mayo (p-Mayo) Score ≥5) or Crohn’s disease (CD; Crohn’s Disease Activity Index (CDAI) >150), requiring VDZ initiation due to prior failure or intolerance to anti-TNF agents or unacceptable side effects from corticosteroids and/or immunosuppressants. Previous analysis of VDZ trough concentrations, presented at United European Gastroenterology Week (UEGW) 2024, demonstrated no significant association between trough levels and clinical response, clinical remission or biomarker dynamics in patients with UC or CD (Barrau et al, personal communication, 2024).
Consequently, VDZ trough concentrations were not included as covariates in the multivariable logistic regression analysis of clinical remission at W52. Anti-αvβ6 antibody measurements were performed using the serum banks, with investigators blinded to clinical, endoscopic or biomarker data. The database was frozen, and only the anti-αvβ6 antibody levels at W0 and W6 were added.
In the longitudinal analysis, five W6 samples from patients with UC were not evaluable. Therefore, anti-αvβ6 antibody levels at W0 and W6 were analysed in 62 patients with paired evaluable samples.
Patients from an independent control cohort (GOLILOR) were included to assess treatment specificity of anti-integrin αvβ6 antibody dynamics. GOLILOR is a prospective, multicentre trial analysing the response to optimisation at W4–W8 in cases of loss of response to golimumab. This cohort comprised patients with UC undergoing golimumab dose intensification for secondary loss of response (defined as a p-Mayo Score >4 with an endoscopic score ≥2) who subsequently underwent golimumab dose escalation from 50 mg every 4 weeks to 100 mg every 4 weeks or from 100 mg every 4 weeks to 100 mg every 2 weeks.17 As with the VEDOPREDIRESP study, anti-αvβ6 antibody measurements were performed with investigators blinded to clinical, endoscopic or biomarker data.
The database was frozen, and only the levels of anti-αvβ6 antibody levels at W0 and W8 were added. The current study evaluated (1) the diagnostic performance of baseline serum anti-integrin αvβ6 autoantibodies in distinguishing UC from CD and their association with disease phenotype; (2) the predictive value of early serum anti-integrin αvβ6 autoantibody titres (baseline and W6 or W8) for sustained clinical remission at W52 in patients with UC; (3) the specificity of early anti-integrin αvβ6 autoantibody titre dynamics to VDZ treatment, as compared with golimumab, in UC; and (4) the association between early anti-integrin αvβ6 autoantibody titre and biomarker and/or endoscopic response at this time.
Clinical outcomes
In the VEDOPREDIRESP cohort, clinical remission at W52 was defined as a p-Mayo Score <3, with no individual subscore >1 for UC and a CDAI <150 for CD among patients treated with VDZ 300 mg every 8 or 4 weeks after receiving standard induction dosing (300 mg intravenously at W0, W2 and W6). At W52, patients were classified into two mutually exclusive groups: those in clinical remission receiving standard VDZ dosing or after dose intensification (300 mg every 4 weeks), and those who discontinued VDZ because of loss of response, adverse events or need for surgery. At W14, faecal calprotectin was analysed for all patients.
Calprotectin remission was defined as faecal calprotectin below 250 µg/g of stool. In the GOLILOR cohort, clinical outcomes following golimumab dose intensification were assessed according to the specific criteria of the original study protocol. Clinical response was defined as a decrease of at least 3 points in the p-Mayo Score, whereas clinical remission was defined as a p-Mayo Score ≤1.
Clinical outcomes were assessed between W2 and W4 and between W4 and W8 after dose intensification, with the latter assessment representing the principal postintensification efficacy assessment.
The patients included in the post hoc UC analysis had active UC (Mayo Clinic Score >4 with an endoscopic score >1) in all cases. In both cohorts, the proportions of patients with Mayo endoscopic score of 3 at baseline were comparable at around 40% in each study. Patients in these two studies were on first-line biologics in 33% and 80% of cases for VDZ and golimumab, respectively.
However, the number of lines of biologic therapy was not associated with the decline in anti-αvβ6 antibodies over time during VDZ treatment. It should be noted, however, that patients on golimumab had failed on golimumab and were therefore on optimised therapy, unlike patients on VDZ, who were VDZ naïve at baseline.
Biomarker measurement
Quantification of IgG anti-integrin αvβ6 autoantibodies
IgG anti-integrin αvβ6 autoantibodies were measured in longitudinally collected serum samples from VEDOPREDIRESP (baseline (W0) and W6) and GOLILOR (W0 and W8) using an in-house quantitative indirect ELISA, adapted from Livanos et al.13 During VDZ induction therapy, five W6 samples from patients with UC were not evaluable. Therefore, anti-αvβ6 antibody levels at W0 and W6 were analysed in 62 patients with paired evaluable samples. All samples were stored at −20°C until analysis.
All measurements were performed with investigators blinded to clinical data. To ensure accurate longitudinal kinetics, baseline and follow-up samples from each individual patient were analysed in the same assay run. Briefly, high-binding 96-well MaxiSorp plates (Thermo Fisher Scientific, Waltham, Massachusetts, USA) were coated with 100 µL/well of recombinant human integrin αvβ6 (R&D Systems, Minneapolis, Minnesota, USA), diluted to 1.5 µg/mL in carbonate-bicarbonate coating buffer for 1 hour, washed three times and blocked for 2 hours.
Serum samples, diluted 1:100, were added (100 µL/well) and incubated for 1 hour. A standard curve was generated using serial dilutions of mouse anti-human αvβ6 monoclonal antibodies (Catalogue No MAB2077Z; MilliporeSigma, Burlington, Massachusetts, USA) from 312.5 ng/mL to 1.22 ng/mL.
After washing, plates were incubated with horseradish peroxidase (HRP)-conjugated secondary antibodies: anti-human IgG-HRP (Catalogue No 31410; Invitrogen, Carlsbad, California, USA), diluted 1:10 000, for samples, and anti-mouse IgG-HRP (Catalogue No 62-6520; Invitrogen), diluted 1:2000, for standards, for 30 min. Plates were then washed, developed with 100 µL/well of 3,3′,5,5′-tetramethylbenzidine (TMB) substrate for 4 min and stopped with 100 µL/well of 2 M sulfuric acid; absorbance was measured at 450 nm using a BioTek microplate reader (BioTek Instruments, Winooski, Vermont, USA), with data acquired using Gen5 software (version 1.1). All incubations were performed at room temperature.
Incubation buffers were sourced from the ELISA Starter Accessory Kit (Bethyl Laboratories, Montgomery, Texas, USA), with 1 mM MgCl2 and CaCl2 added to stabilise the αvβ6 integrin heterodimer, as described by Kuwada et al.10 Anti-integrin αvβ6 antibody titres were interpolated from a four-parameter logistic standard curve (R2=0.99). To define the seropositivity cut-off, we tested this assay on sera from 25 healthy blood donors. The seropositivity threshold was defined as greater than the mean plus 3 SDs of anti-integrin αvβ6 titres measured in these 25 healthy donors (mean age 41.2±12.4 years), corresponding to a value >9.1 ng/mL.
Measurement of serum anti-Saccharomyces cerevisiae antibodies and anti-neutrophil cytoplasmic antibodies
Anti-Saccharomyces cerevisiae antibodies (ASCA; IgA and IgG) and anti-neutrophil cytoplasmic antibodies (ANCA) including anti-myeloperoxidase (anti-MPO) and anti-proteinase 3 (anti-PR3) were measured at baseline (W0) exclusively in the VEDOPREDIRESP cohort. Serum ASCA levels were quantified using an automated fluoroenzyme immunoassay on the Phadia 250 system (Thermo Fisher Scientific). Serum ANCA (anti-MPO and anti-PR3) levels were measured using an automated chemiluminescent immunoassay on the BioFlash system (Inova Diagnostics, San Diego, California, USA).
Measurement of faecal calprotectin
Faecal samples were collected, extracted and diluted to obtain a final dilution of 1:500 using the CALEX Cap device (BÜHLMANN Laboratories AG, Schönenbuch, Switzerland). After a step of centrifugation, the dosage was based on the immunoturbidimetric technique using the fCAL turbo test, adapted on the Optilite instrument (The Binding Site Group, Birmingham, UK).
Statistical analysis
Statistical analyses were performed using GraphPad Prism (version 9.0; GraphPad Software, Boston, Massachusetts, USA). Baseline characteristics were summarised as medians with IQRs for continuous variables and as frequencies with percentages (%) for categorical variables. Group comparisons were conducted using the Mann-Whitney U test for continuous variables and χ2 or Fisher’s exact tests for categorical variables.
Paired comparisons between baseline and follow-up measurements were performed using the Wilcoxon matched-pairs signed-rank test. The diagnostic accuracy of serum anti-integrin αvβ6 antibodies in distinguishing UC from CD was assessed using receiver operating characteristic (ROC) curve analysis, with the optimal threshold determined by Youden’s index. The percentage reduction in serum anti-integrin αvβ6 autoantibody levels from baseline to W6 was evaluated by ROC curve analysis to predict sustained clinical remission.
The optimal percentage reduction was determined using Youden’s index. Multivariable logistic regression was used to identify independent predictors of clinical remission at W52, with results expressed as ORs and 95% CIs. A p value <0.05 was considered statistically significant.
We performed the same analysis to isolate the optimal reduction to predict calprotectin remission at W14.
Results
Cohort characteristics and clinical outcomes (VEDOPREDIRESP cohort)
VEDOPREDIRESP prospectively enrolled 115 adults with IBD, including 67 patients with UC and 48 with CD (table 1). The median age was 39 years (IQR 23–55), with a male-to-female ratio of 0.55. Prior anti-TNF exposure was reported in 80% of participants.
Clinical remission at W52 was achieved in 51% (54% UC, 47% CD). VDZ dose intensification (300 mg every 4 weeks) was required in 45%, of whom 50% achieved remission. Two patients with CD required surgical intestinal resection during follow-up.
Serum anti-integrin αvβ6 antibodies strongly discriminate between UC and CD
Biomarker analysis included 110 paired serum samples (W0 and W6) for anti-integrin αvβ6 and 115 baseline serum samples for ANCA, ASCA and anti-integrin αvβ6. At baseline, serum anti-integrin αvβ6 seropositivity was observed in 94% of patients with UC versus 19% of patients with CD (p<0.001), with median titres of 271 ng/mL (38–884) vs 5.2 ng/mL (2.5–10), respectively (p<0.001) (figure 1A). ROC curve analysis identified 15 ng/mL as the optimal diagnostic threshold (area under the ROC curve (AUC)=0.93, 95% CI 0.88 to 0.98, sensitivity 0.92, specificity 0.95, p<0.001) (figure 1B).
Additional serological biomarker status or levels of ANCA and ASCA did not further improve the high diagnostic accuracy of serum anti-integrin αvβ6 antibodies alone.
Serum anti-integrin αvβ6 levels and IBD phenotype
To assess the relationship between serum anti-integrin αvβ6 and disease phenotype, antibody titres were analysed according to disease extent in UC and disease location and behaviour in CD (figure 2). In UC, titres did not differ significantly by disease extent, with median values of 79 ng/mL (45–275) in patients with proctitis (E1) and 394 ng/mL (184–884) in those with left-sided or extensive colitis (E2/E3) (p=0.140). In CD, titres were significantly higher in patients with colonic or ileocolonic disease (L2–L3) (median 6.3 ng/mL (5.2–12.8)) compared with those with isolated ileal involvement (L1) (median 2.9 ng/mL (1.6–3.4)) (p=0.006).
However, median titres in colonic CD remained >40-fold lower than in UC. No significant differences were observed in titres based on disease behaviour, with median values of 4.06 ng/mL (2.8–9.90) in patients with non-stricturing, non-penetrating disease (B1) and 5.34 ng/mL (3.4–10.3) in those with stricturing or penetrating disease (B2/B3) (p=0.800).
Predictive value of serum anti-integrin αvβ6 dynamics for clinical remission in UC treated with VDZ
At baseline, median serum anti-αvβ6 antibody levels were comparable between patients who achieved clinical remission at W52 and those who did not (250 ng/mL (186–884) vs 225 ng/mL (160–884), p=0.230). Similarly, patients in the highest quartile of baseline αvβ6 titres (>884 ng/mL) did not have higher rates of sustained remission compared with those in lower quartiles (58% vs 45%, p=0.700). In addition, at W6, median serum αvβ6 antibody levels were numerically higher in patients who achieved clinical remission at W52 than in those who did not (239 ng/mL (175–758) vs 207 ng/mL (168–924), p=0.140).
Anti-αvβ6 antibodies at baseline and endoscopic activity in patients with UC (Mayo Score 2 or 3) were similar between the two endoscopic scores. This holds true before starting VDZ or intensification of golimumab (median 232 ng/mL (189–754) vs 243 ng/mL (183–931), p=0.610) for Mayo endoscopic scores 2 and 3, respectively. During VDZ induction therapy, five W6 samples from patients with UC were not evaluable.
Therefore, anti-αvβ6 antibody levels at W0 and W6 were analysed in 62 patients with paired evaluable samples. In this subgroup, the rate of sustained clinical remission at W52 was 61%. Median serum αvβ6 titres in patients with UC decreased significantly from 243 ng/mL (176–884) at W0 to 190 ng/mL (89–320) at W6 (p=0.040) (figure 3A).
A reduction of >30% in αvβ6 levels during this period was identified as the optimal cut-off for predicting sustained remission at W52, with an AUC of 0.79 (95% CI 0.75 to 0.85, p=0.020). This threshold demonstrated a sensitivity of 78%, specificity 80%, positive predictive value 86% and negative predictive value 69% (figure 3B). Neither serum ANCA nor ASCA levels, individually or in combination with αvβ6 titres, improved the prediction of clinical remission at W52 in patients treated with VDZ.
Multivariable logistic regression (table 2) (adjusted for age, disease duration, prior anti-TNF exposure) confirmed that a >30% reduction in serum anti-integrin αvβ6 antibody titres between W0 and W6 was the only independent predictor of sustained clinical remission (OR 5.36, 95% CI 1.12 to 8.54, p=0.040). In the VEDOPREDIRESP study, 22 patients were naïve to biologics before VDZ. In analysing factors associated with anti-αvβ6 autoantibody kinetics using univariable and multivariable analyses, prior anti-TNF exposure was not associated with a >30% reduction in serum anti-integrin αvβ6 antibody titres between W0 and W6 (p=0.360).
Conversely, the association between prior biologic exposure and baseline anti-αvβ6 seropositivity was not evaluated.
Predictive value of serum anti-integrin αvβ6 dynamics for calprotectin normalisation in UC treated with VDZ
Median titres of serum anti-integrin αvβ6 at W0 were similar between patients with UC starting VDZ with faecal calprotectin levels below or above 250 µg/g stools (252 ng/mL (187–825) vs 214 ng/mL (155–932), p=0.230) and decreased similarly between W0 and W6 (−88 ng/mL vs −93 ng/mL, p=0.760). Similarly to clinical remission at W52, a reduction of >30% in αvβ6 levels during this period was a predictor of faecal calprotectin remission at W14, with a sensitivity of 0.74 and specificity of 0.83 (data not shown; these estimates were not included in table 2). We compared calprotectin and antibody levels at baseline (W0) before treatment with VDZ in UC.
These levels were comparable regardless of the calprotectin level at enrolment (correlation coefficient between calprotectin and anti-αvβ6 autoantibody titres in UC at study enrolment in the VEDOPREDIRESP study: r=−0.023). On univariable analysis, faecal calprotectin was associated with sustained clinical remission (p=0.010) but not on multivariable analysis (OR 2.25, 95% CI 0.95–6.32, p=0.056). A significant association between anti-αvβ6 antibodies dynamics at W6 and faecal calprotectin at W14 was observed.
Characteristics and clinical outcomes from the GOLILOR cohort
The GOLILOR cohort included 47 adults with UC who had not previously received an anti-TNF agent other than golimumab in 80% of cases (table 1). After golimumab dose intensification, a clinical response at W4–W8 was observed in 66% of patients escalated to 100 mg every 4 weeks and in 36% of those escalated to 100 mg every 2 weeks. Clinical remission was achieved in 45% of patients at W28, and endoscopic response (Mayo endoscopic score of 0 or 1) was achieved in 41% of patients at W8.
Serum anti-integrin αvβ6 dynamics failed to predict clinical remission in the GOLILOR cohort
Serum anti-integrin αvβ6 antibody levels at baseline were comparable between patients who achieved clinical remission at W4–W8 after golimumab dose intensification and those who did not (median 185 ng/mL (87–363) vs 207 ng/mL (132–492), p=0.570). After golimumab dose intensification, there was no reduction in median serum anti-αvβ6 autoantibody levels in patients with UC, with levels remaining stable from 197 ng/mL (112–368) at W0 to 205 ng/mL (97–457) at W8 (p=0.730) (figure 4). At W8, serum anti-integrin αvβ6 levels did not differ significantly between patients who achieved clinical remission at W4–W8 after golimumab dose intensification and those who did not (median 193 ng/mL (106–321) vs 205 ng/mL (148–381), p=0.770).
Serum anti-integrin αvβ6 at W8 was not associated with endoscopic response in the GOLILOR cohort
We investigated a correlation between the endoscopic response at W8 after golimumab optimisation, as reported by the Mayo endoscopic score, and anti-αvβ6 antibody levels. At W8, 41% of patients were in endoscopic response (Mayo endoscopic score 0 or 1) and 31% in endoscopic remission (Mayo endoscopic score 0). Median titres of serum anti-integrin αvβ6 at W8 were similar between patients with or without endoscopic response (102 ng/mL vs 123 ng/mL, p=0.570).
These results were similar when we analysed median titres of serum anti-integrin αvβ6 at W8 between patients with or without endoscopic remission (106 ng/mL vs 119 ng/mL, p=0.630).
Decrease of > 30% in anti-integrin αvβ6 was significantly associated with normalisation of calprotectin in both univariable and multivariable analyses
We first performed univariable and then multivariable analyses to identify the factors associated with early normalisation (W14) of faecal calprotectin in patients with UC treated with VDZ (table 3). Only a reduction of >30% in anti-αvβ6 antibodies was significantly associated with normalisation of calprotectin in both univariable and multivariable analyses. In this analysis, the lack of statistical power means that, on the one hand, we cannot rule out other associated factors, and on the other hand, we could not reliably assess the increase in anti-αvβ6 antibody levels in the event of a decline in kinetics due to very wide CIs.
We also analysed factors associated with normalisation of faecal calprotectin as recommended on STRIDE-II (150 µg/g of stool) at W14 (online supplemental table 1). Only a reduction of >30% in anti-αvβ6 antibodies was associated with faecal calprotectin normalisation (OR 2.54. 95% CI 1.02–12.24; p=0.048).
Anti-integrin αvβ6 levels are not associated with VDZ levels
We investigated an association between anti-integrin αvβ6 levels at W6 and VDZ levels at W6. No correlation could be demonstrated. The mean VDZ levels at W6 and W14 were comparable between the groups, regardless of whether or not there was a decrease in anti-integrin αvβ6 levels.
Mean VDZ levels at W6 according to whether or not anti-integrin αvβ6 levels had decreased were respectively 22.9±9.2 µg/mL vs 23.4±7.9 µg/mL (p=0.430). Mean VDZ levels at W14 according to whether or not anti-integrin αvβ6 levels decreased were respectively 18.09±8.12 µg/mL vs 18.8±9.41 µg/mL (p=0.540).
Furthermore, we investigated the factors associated with a decrease in these levels between W0 and W6 under VDZ using logistic regression. In our study, no demographic, patient or disease-related parameters were predictive of these kinetics. In particular, the severity of initial UC (total Mayo Clinic Score), endoscopic activity and the number of prior biologic therapies were not associated with this kinetics (table 3).
Discussion
Anti-integrin αvβ6 autoantibodies have emerged as a highly accurate serological biomarker for UC. A recent meta-analysis of six independent cohorts confirmed their strong diagnostic accuracy with a pooled sensitivity of 83% and specificity of 93%.18 However, specificity was shown to vary according to patient age, geographical region and control group composition, particularly when CD was used as the comparator. In the present study, through a post hoc analysis of an independent French multicentre cohort from a geographically distinct population, these findings were fully replicated and further extended.
Using an optimal cut-off of 15 ng/mL, anti-αvβ6 autoantibodies achieved 92% sensitivity and 95% specificity for discriminating UC from CD. The addition of conventional serological markers, including perinuclear ANCA (pANCA) and ASCA, did not improve diagnostic performance. Collectively, these results support serum anti-integrin αvβ6 autoantibodies as a robust, stand-alone serological biomarker with superior diagnostic accuracy compared with traditional pANCA and ASCA testing for the diagnosis of UC.
Multi-omics studies have convincingly positioned colonic CD on a biological continuum between ileal CD and UC, with genetic risk scores,19 serum proteomics,20 faecal metabolomics, metaproteomics and microbiota composition,21 all showing greater molecular similarity between colonic CD and UC than between colonic and ileal CD. Pertsinidou et al,12 who reported elevated serum anti-αvβ6 autoantibody levels in colonic CD compared with healthy controls, although statistical significance was not achieved in their validation cohort, proposed that these autoantibodies may reflect a distinct immunological phenotype shared by both conditions rather than a mere epiphenomenon of colonic inflammation. Our phenotype-stratified analyses showed that anti-αvβ6 autoantibody titres in UC were consistently elevated irrespective of disease extent.
In CD, titres were significantly higher in patients with colonic or ileocolonic involvement (L2/L3; median 6.3 ng/mL, IQR 5.2–12.8) than in those with isolated ileal disease (p=0.006). However, median titres in colonic CD remained >40-fold lower than in UC. This marked quantitative dissociation highlights the distinctive immunobiology of UC within the IBD spectrum and further reinforces the diagnostic utility of anti-αvβ6 autoantibodies for this condition.
Although multi-omics data demonstrate a molecular continuum across IBD, our findings indicate that serum anti-αvβ6 autoantibodies constitute a defining serological feature of UC.
Beyond its diagnostic superiority, serial serum anti-integrin αvβ6 autoantibody measurement revealed a pharmacodynamic role in VDZ-treated patients with UC. Although baseline levels did not differ between responders and non-responders, an early reduction exceeding 30% from baseline to W6 strongly predicted sustained W52 remission. Moreover, this early reduction exceeding 30% from baseline to W6 was a good predictor of early calprotectin normalisation at W14.
Remarkably, lower baseline serum anti-αvβ6 levels also predicted clinical remission with carotegrast methyl, an oral α4-integrin antagonist, in the ASPECT study.22 Thus, across two distinct α4-integrin-targeting therapies (monoclonal antibody and oral small molecule), serum anti-αvβ6 autoantibody titres and their dynamics show consistent associations with subsequent clinical and biomarker remission.
However, our analysis of the VEDOPREDIRESP cohort alone could not resolve whether the pronounced early decline in serum anti-αvβ6 autoantibody titres observed among VDZ responders simply mirrored successful suppression of mucosal inflammation or reflected a direct pharmacodynamic consequence of α4β7 blockade. To address this, we analysed an independent cohort of patients with UC treated with a mechanistically distinct agent—golimumab. Notably, serum anti-αvβ6 titres remained unchanged in patients treated with golimumab from W0 to W8 (p=0.730).
Moreover, we did not report any relationship between serum anti-αvβ6 titres from W8 and endoscopic response at the same time after golimumab intensification. This observation suggests that systemic TNF blockade may not materially affect serum αvβ6 antibody levels and may support the specificity of the dynamic changes observed with gut-selective integrin-targeted therapy.
Conversely to some studies, we did not report any correlation between anti-αvβ6 autoantibody titres and faecal calprotectin or Mayo endoscopic score. In a recent study, Pertsinidou et al12 showed that higher autoantibody levels were linked to an aggressive disease course, remaining stable in aggressive UC but decreasing in indolent disease (p=0.003). However, we analysed only data about VDZ response at 1 year or golimumab response at 26 weeks.
In another study,16 the authors showed that patients with Mayo endoscopic subscore of 3 had significantly higher anti-αvβ6 autoantibody titres than those with endoscopic remission. In our prospective and multicentre study using golimumab intensification, at W8, anti-αvβ6 autoantibody titres were not significantly associated with the Mayo endoscopic score. The decline in anti-αvβ6 autoantibody levels during VDZ treatment was not clearly associated with clinical or endoscopic disease activity but may reflect an early pharmacodynamic effect of VDZ.
Recent single-cell and repertoire studies have localised the IgG-skewed B-cell response in UC predominantly to the inflamed colonic mucosa, showing expansion of β7-integrin+, IgG1-enriched plasmablasts, plasma cells with reduced somatic hypermutation and, in one instance, an αvβ6-autoreactive clone.23 These findings support the concept that anti-αvβ6 antibodies can arise within a broader, type 1-polarised, gut-homing IgG response. In this context, our observations are consistent but do not prove the hypothesis that circulating anti-αvβ6 autoantibodies might reflect, at least in part, a dysregulated mucosal B-cell response, and that their treatment-specific dynamics may relate to differential effects of integrin- versus TNF-targeted therapies on α4β7-dependent B-cell trafficking, rather than purely non-specific control of inflammation. Prospective mechanistic dedicated studies, integrating serial anti-αvβ6 measurements with longitudinal mucosal and peripheral immune profiling across different treatment classes, will be required to better understand to what extent these autoantibodies and their dynamics are directly pathogenic, secondary epiphenomena of tissue injury or primarily useful as treatment-linked pharmacodynamic and prognostic biomarkers.
A central unanswered question is whether anti-αvβ6 antibodies are potent biomarkers of epithelial injury or active contributors to disease pathogenesis. Experimental data suggest that antibody binding to αvβ6 integrin can interfere with its interaction with extracellular matrix ligands, such as fibronectin, and impair transforming growth factor-β (TGF-β) activation. Such interference could compromise epithelial repair mechanisms, perpetuate barrier dysfunction and amplify mucosal and biliary inflammation.
Anti-αvβ6 antibodies might participate in a self-reinforcing cycle of epithelial injury and immune activation.24
Our study has several limitations that warrant consideration. Although serum samples were prospectively collected, the post hoc design introduces potential selection bias. In the VEDOPREDIRESP design, outcome was defined by clinical activity score alone without data of biomarker or endoscopic score.
The modest sample size limited the statistical power to detect differences in rare phenotypes and precluded robust subgroup analyses. Performance of the biomarker to predict biochemical and endoscopic remission at W52 was not analysed. Moreover, the present VEDOPREDIRESP cohort was not specifically composed of newly diagnosed patients with IBD, and the diagnostic performance of anti-αvβ6 antibodies was not evaluated specifically in a newly diagnosed or inception population.
Finally, the association between early antibody decline and W52 remission was not adjusted for subsequent VDZ treatment intensification and may therefore be affected by post-baseline treatment optimisation.
However, our study has several strengths. It is a study reporting kinetics of serum anti-αvβ6 titres during the early induction phase of treatment. Furthermore, we analysed objective assessments of disease activity, including endoscopic outcomes for golimumab and faecal calprotectin for VDZ.
The threshold for a decrease in anti-αvβ6 antibody levels identified in our cohort remains exploratory and should not be used to guide treatment decisions in clinical practice. Prospective external validation in independent cohorts will be required to confirm its reproducibility, predictive performance and clinical utility.
In conclusion, our findings suggest that early changes in serum anti-αvβ6 autoantibody titres may serve as a relevant VDZ-associated pharmacodynamic biomarker in UC, with potential predictive value for biomarker remission after induction and long-term remission within the first weeks of therapy. Although the decline in anti-αvβ6 antibody was associated with subsequent clinical remission, the biological basis of this association remains uncertain. It may reflect a reduction in the underlying mucosal autoimmune response and/or restoration of epithelial homeostasis during effective treatment.
Such a non-invasive marker could potentially support treat-to-target strategies, guide timely treatment adjustments and improve the cost-effectiveness of care. Analysis of the kinetics of anti-αvβ6 (W0 vs W6) may help identify, with a high degree of predictability, patients who will respond to long-term VDZ treatment for UC. In the absence of a significant early decline in these antibodies, a proactive strategy could be proposed for these patients.
Validation in larger, prospective, dedicated and multiethnic cohorts will be essential before these measurements can be incorporated into routine clinical practice.
Supplementary material