Soluble Mucosal Addressin Cell Adhesion Molecule 1 and Retinoic Acid are Potential Tools for Therapeutic Drug Monitoring in Patients with Inflammatory Bowel Disease Treated with Vedolizumab: A Proof of Concept Study

Soluble Mucosal Addressin Cell Adhesion Molecule 1 and Retinoic Acid are Potential Tools for... Abstract Background and Aims Vedolizumab [VDZ], a humanized monoclonal antibody targeting α4β7 integrin, is effective in induction and maintenance therapy in patients with inflammatory bowel disease [IBD] who have not adequately responded to standard therapies, and high vedolizumab trough levels [VTLs] have been associated with clinical remission. The α4β7 integrin binds to endothelial MAdCAM-1 and is upregulated by retinoic acid [RA]. The aim of this study was to determine the relationships between soluble MAdCAM-1 [sMAdCAM-1] and RA concentrations during clinical remission with VDZ maintenance therapy. Methods In a retrospective study performed in IBD patients treated with VDZ, we measured VTL, sMAdCAM-1 and RA concentrations. Results Among the 62 included patients [38 Crohn’s disease], 24 relapsed and 38 stayed in remission from Weeks 10 to 30 after VDZ initiation. During this maintenance therapy, the median values of VTLs and RA were 15.4 µg/mL and 0.97 ng/mL, respectively, whereas sMAdCAM-1 was undetectable [<0.41 ng/mL] in 67.3% of samples. The positive predictive value [PPV] of undetectable sMAdCAM-1 for clinical remission was 80.0%, with a corresponding sensitivity of 74.6%. On multivariate analysis, undetectable sMAdCAM-1 and high VTLs [>19 µg/mL] were independently associated with clinical remission [OR = 7.5, p = 0.006 and OR = 2.2, p = 0.045, respectively]. The combination of sMAdCAM-1 < 0.41 ng/mL and VTL > 19 µg/mL was the best pharmacokinetic profile, with a PPV of 95.2%. Median values of sMAdCAM-1 and RA were significantly higher [p = 0.0001] before VDZ therapy than during the follow-up [sMAdCAM-1: 40.5 vs < 0.41 ng/mL; RA: 1.7 vs 0.97 ng/mL]. Only RA > 1.86 ng/mL before VDZ therapy was predictive of clinical remission during the follow-up (Area Under a Receiver Operating Characteristic curve [AUROC] = 80.7%). Conclusions Undetectable sMAdCAM-1 appears strongly associated with clinical remission during VDZ maintenance therapy. Combination of undetectable sMAdCAM-1 with high VTL is also potentially interesting for therapeutic drug monitoring. Baseline RA concentrations are predictive of clinical remission. These findings need to be confirmed in further prospective studies. Soluble MAdCAM-1, retinoic acid, vedolizumab, inflammatory bowel disease, therapeutic drug monitoring 1. Introduction In recent decades, biologic therapies such as anti-tumor necrosis factor [TNF] and vedolizumab [VDZ] have been developed for treating patients with inflammatory bowel disease [IBD].1–4 To optimize their use, therapeutic drug monitoring [TDM] has emerged as an interesting approach. Performances of TDM have been so far assessed with anti-TNF agents, and several decision trees based on trough levels of anti-TNF and anti-drug antibodies have been proposed.5–7 VDZ is a humanized monoclonal antibody targeting α4β7 integrin. It modulates inflammation of the gastrointestinal wall by limiting lymphocyte homing to the lamina propria without inducing systemic immunosuppression. Vedolizumab was approved in 2014 for induction and maintenance therapy in IBD patients who had not adequately responded to one or more standard therapies [corticosteroids, immunomodulators, or anti-TNF]. We recently reported the association between low VDZ trough levels [VTLs] during induction therapy and the need for additional doses within 6 months.8 Other recent studies have reported the association of high VTLs with clinical remission.9–12 However, other factors may be involved in the clearance of the drug and in clinical outcomes. To the best of our knowledge, none of these factors has been studied so far. Some authors evaluated the integrin α4β7 expression on multiple lymphocyte subsets before and during VDZ therapy, but the method used [flow cytometry] appears hardly applicable in clinical practice.12 Leukocyte recruitment is pivotal for the initiation and perpetuation of IBD and is controlled by the interactions of chemokines and adhesion molecules. Mucosal addressin cell adhesion molecule 1 [MAdCAM-1] is a cell-surface immunoglobulin superfamily member that promotes the adhesion of T and B cells to vascular endothelium and is critical for lymphocyte homing to the gut.13 Its expression is increased in patients with IBD and in animal colitis models.14,15 In a recent placebo-controlled trial evaluating an anti-MAdCAM-1 antibody, Vermeire et al. reported a correlation between the circulating soluble form of MAdCAM-1 [sMAdCAM-1] and its tissue expression.16 Vitamin A is the term given to a collection of related molecules known as retinoic acid [RA].17 During vitamin A metabolism, the irreversible conversion of retinal to RA is catalyzed by retinaldehyde dehydrogenase [RALDH], which is expressed by dendritic cells [DCs] from Peyer patches and mesenteric lymph nodes.18 The RALDH allows intestinal DCs to convert retinal to RA, which in turn induces the expression of the gut homing receptor α4β7 on T cells.18–20 Upregulation of α4β7 by RA on the surface of B cells has also been demonstrated.21 Polymorphism in the CYP26B1 enzyme is likely involved in the degradation of RA and has been described in patients with Crohn’s disease [CD], supporting the role of vitamin A in the pathophysiology of CD.22 In this study, we retrospectively examined the relationships between sMAdCAM-1 and RA concentrations during the response to VDZ in IBD patients. We hypothesized that patients with high levels of circulating RA could be potential good responders to VDZ, because RA induces the lymphocyte expression of α4β7, which is the target of VDZ. In contrast, in the presence of high levels of sMAdCAM-1, the α4β7 integrin expressed at the lymphocyte surface could be overloaded with sMAdCAM-1, reducing the accessibility of the integrin to VDZ, and this could therefore be associated with a lack of response to VDZ. 2. Methods 2.1. Patients All consecutive IBD patients from three referral centers who were treated with VDZ as maintenance therapy and in clinical remission after induction therapy were retrospectively included in the present study. They had failed or had developed intolerance to at least one line of anti-TNF. Three hundred milligrams of VDZ were administered at Weeks [W] 0, 2, and 6 as induction therapy, and then were administered once every 8 weeks during the maintenance therapy. Serum samples during the induction phase were not collected. At W10, clinical activity was assessed, and patients in whom clinical remission was not achieved received an additional dose of 300 mg VDZ at W10 followed by an infusion of VDZ once every 4 weeks as optimization of the treatment during the maintenance therapy. Clinical activity was assessed at W10 in all patients, at W30 in patients in remission, at W14 in optimized patients and in patients who relapsed before W30 (using the Harvey–Bradshaw index [HBI] and faecal calprotectin [fCal] in patients with CD, and the Mayo score in patients with ulcerative colitis [UC]). Clinical remission was defined as a HBI of ≤3 with fCal levels of <250 µg/g stools for CD, and as a Mayo score of ≤3 with an endoscopic subscore of 0 to 1 for UC without concomitant corticosteroids. Primary non-response to VDZ at W14 and relapse were defined by a HBI of >4 with fCal of >250 µg/g stools for CD, and by a Mayo score of >4 with an endoscopic subscore of >1 for UC. Patients with primary non-response to VDZ at W14 in spite of VDZ infusion at W10 were excluded from our study, which focused on maintenance therapy. A medical visit was performed before each VDZ infusion. In all patients, sMAdCAM-1 and RA concentrations were measured in the blood before the first VDZ infusion. In patients in clinical remission, measurements of VTL, sMAdCAM-1, and RA concentrations were performed at W30. For patients in clinical relapse after W10, measurements of VTL, sMAdCAM-1, and RA concentrations were performed at the time of relapse. Measurements of VDZ, sMAdCAM-1, and RA concentrations were performed retrospectively. Hence, the management of patients was independent of these results. All patients signed an informed written consent to the protocol, which was approved by the Ethics Committee of Saint-Etienne University and Centre National Informatique et Liberté [CNIL 1849323 v 0]. 2.2. Measurements of vedolizumab, soluble MAdCAM-1, and retinoic acid concentrations Human sMAdCAM-1 concentrations were measured using the HycultBiotech ELISA assay [Clinisciences, Montrouge, France]. Briefly, the human sMAdCAM-1 ELISA is a ready to use solid-phase sandwich ELISA that allows the determination of sMAdCAM-1 concentrations between 0.41 and 100 ng/mL. We tested the spike of VDZ at a range of between 1 and 30 µg/ml in a V0 sample with a high concentration of sMAdCAM-1 [<5%]. We did not observe any significant differences in sMAdCAM-1 with or without VDZ. Moreover, the situation was exactly the same when dosing withVDZ in the presence of sMAdCAM-1. Again, we did not observe any interference in VDZ measurement in the presence of sMAdCAM-1, as in the V0 samples. Human RA concentrations were measured with the CUSABIO competitive inhibition ELISA assay [Clinisciences, Montrouge, France]. The detection range of this assay is from 0.42 to 10 ng/mL. Vedolizumab and anti-VDZ antibodies were measured using the Lisa-Tracker duo VDZ ELISA assay [Theradiag, Marne-la-Vallée, France]. The assay ranges from 2 to 60 µg/mL for VDZ and from 35 to 500 ng/mL for anti-VDZ antibodies. We used in this study a drug-sensitive ELISA assay for monitoring anti-VDZ antibodies. This assay has been designed to reduce the formation of complexes between VDZ and anti-VDZ antibodies by using specific buffers. However, with the drug-sensitive ELISA technique, when VDZ is detectable in the serum, the value of the anti-VDZ antibody is considered ‘inconclusive’. 2.3. Statistical analysis The primary objective was to evaluate the sMAdCAM-1 and RA concentration, in addition to the VTL, and to assess clinical remission under VDZ maintenance therapy. First, a cross-sectional analysis was carried out including only measurements of sMAdCAM-1, RA, and VTL [performed during maintenance therapy at W30 or at the time of relapse]. Linear correlations between VDZ, sMAdCAM-1, or RA concentrations were assessed using the Pearson correlation coefficient. The most associated threshold with clinical remission for each biological variable were identified by receiver operating characteristic [ROC] curves. Factors associated with clinical remission were then identified using uni- and multivariate logistic regressions, including dichotomized variables defined by previously identified thresholds. Variables that achieved a p < 0.1 value on univariate analysis were included in the multivariate analysis. Pharmacokinetic profiles combining dichotomized values significantly associated with clinical remission on multivariate analysis were then tested according to proportions of patients in remission or relapse, and were compared with the Chi2 or Fisher tests, as appropriate. Similar analyses focusing on quartiles of the biomarkers were also performed. The one-sided Cochrane–Armitage trend test was used to compare quartiles. Second, in a longitudinal analysis, the median concentrations of sMAdCAM-1 and RA before and during VDZ maintenance therapy were compared using the paired Mann–Whitney test. Predictive performances for clinical remission of these biomarkers (sensitivity, specificity, accuracy, positive predictive value [PPV], and negative predictive value [NPV]) were tested by ROC curves. Statistical analysis was performed using IBM SPSS 20.0.0 [IBM, Somers, NY]. The significance level was defined as p ≤ 0.05. 3. Results 3.1. Cross-sectional analysis Characteristics of the population study are reported in Table 1. Among the 62 patients in clinical remission at W14 [38 CD] considered for analysis, ~50% were female, and the median age was 38 years [median; IQR25–75: 30–56]. All were previously treated with immunosuppressive drugs [azathioprine or methotrexate] and usually two lines of anti-TNF agents. The median interval between anti-TNF discontinuation and VDZ initiation was 4 weeks [IQR25–75: 4–6] and 8 weeks [IQR25–75: 8–12] for adalimumab and infliximab, respectively. Residual trough levels of anti-TNF were probably low in this study and we did not expect a strong carryover effect. From W10 to W30, 24 patients relapsed, whereas 38 stayed in clinical remission [Figure 1]. Sampling was performed at the time of relapse in patients who relapsed and was therefore not at trough time in all the 24 patients. Sampling was not performed during the induction phase. However, in these cases, sampling was performed at least 4 weeks after the last infusion. Conversely, sampling was performed 8 weeks after the last infusion in 28 patients in clinical remission with no optimization, and 4 weeks after the last infusion in 12 patients optimized and in clinical remission. A total of 124 assays were performed, with 62 during VDZ maintenance therapy and 62 before the first infusion of VDZ. During maintenance therapy, 38 determinations of VTL, sMAdCAM-1, and RA concentrations were performed from patients in clinical remission and 24 from patients who relapsed [Figure 1]. Table 1. Characteristics of the population. Population study [n = 62] Women, n [%] 32 [51.6] Age year, median [IQR25–75] 38 [30–56] Weight kg, median [IQR25–75] 63 [58–74] Crohn,n [%] 38 [61.3] Age at diagnosis  <17 years [A1] 6 [15.8]  17–40 years [A2] 24 [63.2]  above 40 years [A3] 8 [21.0] Behaviour*, n [%]  non-stricturing, non-penetrating [B1] 17 [44.7]  stricturing [B2] 17 [44.7]  penetrating [B3] 4 [10.6] Location*, n [%]  ileal [L1] 13 [34.2]  colonic [L2] 4 [10.5]  ileocolonic [L3] 21 [55.3]  perianal disease 7 [18.4] Ulcerative colitis, n [%]* 24 [38.7]  proctitis [E1] 2 [8.3]  left sided [E2] 4 [16.7]  extensive [E3] 18 [75.0] Previous medication, n [%] Azathioprine 48 [77.4] Methotrexate 14 [22.6] Anti-TNF 62 [100]  1 line anti-TNF 11 [17.7]  2 lines anti-TNF 51 [82.3] History of surgery for IBD,n [%] 27 [43.5] Population study [n = 62] Women, n [%] 32 [51.6] Age year, median [IQR25–75] 38 [30–56] Weight kg, median [IQR25–75] 63 [58–74] Crohn,n [%] 38 [61.3] Age at diagnosis  <17 years [A1] 6 [15.8]  17–40 years [A2] 24 [63.2]  above 40 years [A3] 8 [21.0] Behaviour*, n [%]  non-stricturing, non-penetrating [B1] 17 [44.7]  stricturing [B2] 17 [44.7]  penetrating [B3] 4 [10.6] Location*, n [%]  ileal [L1] 13 [34.2]  colonic [L2] 4 [10.5]  ileocolonic [L3] 21 [55.3]  perianal disease 7 [18.4] Ulcerative colitis, n [%]* 24 [38.7]  proctitis [E1] 2 [8.3]  left sided [E2] 4 [16.7]  extensive [E3] 18 [75.0] Previous medication, n [%] Azathioprine 48 [77.4] Methotrexate 14 [22.6] Anti-TNF 62 [100]  1 line anti-TNF 11 [17.7]  2 lines anti-TNF 51 [82.3] History of surgery for IBD,n [%] 27 [43.5] IQR: interquartile range; TNF: tumor necrosis factor; *according to the Montreal Classification. View Large Table 1. Characteristics of the population. Population study [n = 62] Women, n [%] 32 [51.6] Age year, median [IQR25–75] 38 [30–56] Weight kg, median [IQR25–75] 63 [58–74] Crohn,n [%] 38 [61.3] Age at diagnosis  <17 years [A1] 6 [15.8]  17–40 years [A2] 24 [63.2]  above 40 years [A3] 8 [21.0] Behaviour*, n [%]  non-stricturing, non-penetrating [B1] 17 [44.7]  stricturing [B2] 17 [44.7]  penetrating [B3] 4 [10.6] Location*, n [%]  ileal [L1] 13 [34.2]  colonic [L2] 4 [10.5]  ileocolonic [L3] 21 [55.3]  perianal disease 7 [18.4] Ulcerative colitis, n [%]* 24 [38.7]  proctitis [E1] 2 [8.3]  left sided [E2] 4 [16.7]  extensive [E3] 18 [75.0] Previous medication, n [%] Azathioprine 48 [77.4] Methotrexate 14 [22.6] Anti-TNF 62 [100]  1 line anti-TNF 11 [17.7]  2 lines anti-TNF 51 [82.3] History of surgery for IBD,n [%] 27 [43.5] Population study [n = 62] Women, n [%] 32 [51.6] Age year, median [IQR25–75] 38 [30–56] Weight kg, median [IQR25–75] 63 [58–74] Crohn,n [%] 38 [61.3] Age at diagnosis  <17 years [A1] 6 [15.8]  17–40 years [A2] 24 [63.2]  above 40 years [A3] 8 [21.0] Behaviour*, n [%]  non-stricturing, non-penetrating [B1] 17 [44.7]  stricturing [B2] 17 [44.7]  penetrating [B3] 4 [10.6] Location*, n [%]  ileal [L1] 13 [34.2]  colonic [L2] 4 [10.5]  ileocolonic [L3] 21 [55.3]  perianal disease 7 [18.4] Ulcerative colitis, n [%]* 24 [38.7]  proctitis [E1] 2 [8.3]  left sided [E2] 4 [16.7]  extensive [E3] 18 [75.0] Previous medication, n [%] Azathioprine 48 [77.4] Methotrexate 14 [22.6] Anti-TNF 62 [100]  1 line anti-TNF 11 [17.7]  2 lines anti-TNF 51 [82.3] History of surgery for IBD,n [%] 27 [43.5] IQR: interquartile range; TNF: tumor necrosis factor; *according to the Montreal Classification. View Large Figure 1. View largeDownload slide Description of the study design. VTLs: vedolizumab trough levels; sMAdCAM-1: soluble mucosal addressin cell adhesion molecule 1; CD: Crohn’s disease. Figure 1. View largeDownload slide Description of the study design. VTLs: vedolizumab trough levels; sMAdCAM-1: soluble mucosal addressin cell adhesion molecule 1; CD: Crohn’s disease. During maintenance therapy, the median VTL was 15.4 µg/mL [IQR25–75: 7.7–26.5] and there was no anti-VDZ antibody. Median sMAdCAM-1 was significantly lower [p < 0.01] in patients who stayed in remission compared with in those who relapsed [Figure 2]. Soluble MAdCAM-1 was undetectable [< 0.41 ng/mL] in 67.3% of samples [IQR25–75: 0.00–1.29], and the highest level of sMAdCAM-1 was 16.54 ng/mL; it was undetectable in 87.4% of patients in clinical remission compared with in 45.1% of relapsing patients [p = 0.003]. The median RA concentration was 0.97 ng/mL [IQR25–75: 0.42–1.10], and RA was undetectable [< 0.42 ng/mL] in 25.2% of samples. There was no linear correlation between these biomarkers [Pearson’s correlation coefficient: 0.20 and – 0.07 between VTL and sMAdCAM-1 and between VTL and RA, respectively; –0.10 between sMAdCAM-1 and RA]. Figure 2. View largeDownload slide sMAdCAM-1 [A], retinoic acid [B] and vedolizumab trough level [C] concentrations in patients who relapsed or who stayed in remission during vedolizumab maintenance therapy. sMAdCAM-1: soluble mucosal addressin cell adhesion molecule 1. The box plots show median, upper, and lower quartiles of the data; the whiskers indicate the 95% confidence interval of the values. Median and interquartile ranges [25–75] are indicated below the graphs. Figure 2. View largeDownload slide sMAdCAM-1 [A], retinoic acid [B] and vedolizumab trough level [C] concentrations in patients who relapsed or who stayed in remission during vedolizumab maintenance therapy. sMAdCAM-1: soluble mucosal addressin cell adhesion molecule 1. The box plots show median, upper, and lower quartiles of the data; the whiskers indicate the 95% confidence interval of the values. Median and interquartile ranges [25–75] are indicated below the graphs. The optimal thresholds of biomarkers associated with clinical remission with their diagnostic performances were determined by ROC curves and reported in Table 2. The most accurate biomarker was undetectable sMAdCAM-1 [threshold < 0.41 ng/mL]. On univariate analysis by logistic regression [Table 3], undetectable sMAdCAM-1, age > 38 years, RA < 1.05 ng/mL, VTL > 19 µg/mL, type of IBD [UC vs CD], and optimization of the treatment were associated with clinical remission during the follow-up and were included on the multivariate analysis. On multivariate analysis, only undetectable sMAdCAM-1 and VTL > 19 µg/mL remained independently associated with clinical remission [OR = 7.5, 95% CI: 1.3–25.9, p = 0.006 and OR = 2.2, 95% CI: 1.1–30.5, p = 0.045, respectively], with a trend toward a significant association for RA < 1.05 ng/mL [p = 0.062]. Table 2. Identification of optimal thresholds associated with clinical remission for vedolizumab trough levels, sMAdCAM-1 and retinoic acid by receiver operating characteristic [ROC] curves during vedolizumab maintenance therapy. Threshold Sensitivity Specificity Accuracy PPV NPV Vedolizumab trough level >19.0 µg/mL 47.5 65.0 0.50 72.1 39.7 sMAdCAM-1 <0.41 ng/mL [undetectable] 74.6 54.2 0.69 80.0 46.4 Retinoic acid <1.05 ng/mL 59.0 65.6 0.57 78.3 52.0 Threshold Sensitivity Specificity Accuracy PPV NPV Vedolizumab trough level >19.0 µg/mL 47.5 65.0 0.50 72.1 39.7 sMAdCAM-1 <0.41 ng/mL [undetectable] 74.6 54.2 0.69 80.0 46.4 Retinoic acid <1.05 ng/mL 59.0 65.6 0.57 78.3 52.0 sMAdCAM-1: soluble mucosal addressin cell adhesion molecule-1; NPV: negative predictive value; PPV: positive predictive value. View Large Table 2. Identification of optimal thresholds associated with clinical remission for vedolizumab trough levels, sMAdCAM-1 and retinoic acid by receiver operating characteristic [ROC] curves during vedolizumab maintenance therapy. Threshold Sensitivity Specificity Accuracy PPV NPV Vedolizumab trough level >19.0 µg/mL 47.5 65.0 0.50 72.1 39.7 sMAdCAM-1 <0.41 ng/mL [undetectable] 74.6 54.2 0.69 80.0 46.4 Retinoic acid <1.05 ng/mL 59.0 65.6 0.57 78.3 52.0 Threshold Sensitivity Specificity Accuracy PPV NPV Vedolizumab trough level >19.0 µg/mL 47.5 65.0 0.50 72.1 39.7 sMAdCAM-1 <0.41 ng/mL [undetectable] 74.6 54.2 0.69 80.0 46.4 Retinoic acid <1.05 ng/mL 59.0 65.6 0.57 78.3 52.0 sMAdCAM-1: soluble mucosal addressin cell adhesion molecule-1; NPV: negative predictive value; PPV: positive predictive value. View Large Table 3. Uni- and multivariate analysis by logistic regression to identify associated factors with clinical remission during vedolizumab maintenance therapy. Univariate analysis Multivariate analysis p value OR 95% CI p value OR 95% CI sMAdCAM-1 <0.41 ng/mL [undetectable] 0.004 6.5 1.3–19.4 0.006 7.5 1.3–25.9 Retinoic acid <1.05 ng/mL 0.045 2.3 1.1–14.5 0.062 2.5 0.9–40.3 Vedolizumab trough level >19 µg/mL 0.061 2.7 1.0–20.4 0.045 2.2 1.1–30.5 Age >38 years 0.025 3.1 1.2–8.7 0.136 2.3 0.8–6.9 Weight >63 kg 0.60 1.4 0.5–3.9 Non-optimized vs optimized therapy 0.083 3.1 0.9–12.7 0.240 1.9 0.5–12.6 Ulcerative colitis vs Crohn’s disease 0.078 6.7 0.9–69.7 0.200 0.5 0.1–1.4 Univariate analysis Multivariate analysis p value OR 95% CI p value OR 95% CI sMAdCAM-1 <0.41 ng/mL [undetectable] 0.004 6.5 1.3–19.4 0.006 7.5 1.3–25.9 Retinoic acid <1.05 ng/mL 0.045 2.3 1.1–14.5 0.062 2.5 0.9–40.3 Vedolizumab trough level >19 µg/mL 0.061 2.7 1.0–20.4 0.045 2.2 1.1–30.5 Age >38 years 0.025 3.1 1.2–8.7 0.136 2.3 0.8–6.9 Weight >63 kg 0.60 1.4 0.5–3.9 Non-optimized vs optimized therapy 0.083 3.1 0.9–12.7 0.240 1.9 0.5–12.6 Ulcerative colitis vs Crohn’s disease 0.078 6.7 0.9–69.7 0.200 0.5 0.1–1.4 CI: confidence interval; OR: odds ratio; sMAdCAM-1: soluble mucosal addressin cell adhesion molecule-1: TNF: tumor necrosis factor. View Large Table 3. Uni- and multivariate analysis by logistic regression to identify associated factors with clinical remission during vedolizumab maintenance therapy. Univariate analysis Multivariate analysis p value OR 95% CI p value OR 95% CI sMAdCAM-1 <0.41 ng/mL [undetectable] 0.004 6.5 1.3–19.4 0.006 7.5 1.3–25.9 Retinoic acid <1.05 ng/mL 0.045 2.3 1.1–14.5 0.062 2.5 0.9–40.3 Vedolizumab trough level >19 µg/mL 0.061 2.7 1.0–20.4 0.045 2.2 1.1–30.5 Age >38 years 0.025 3.1 1.2–8.7 0.136 2.3 0.8–6.9 Weight >63 kg 0.60 1.4 0.5–3.9 Non-optimized vs optimized therapy 0.083 3.1 0.9–12.7 0.240 1.9 0.5–12.6 Ulcerative colitis vs Crohn’s disease 0.078 6.7 0.9–69.7 0.200 0.5 0.1–1.4 Univariate analysis Multivariate analysis p value OR 95% CI p value OR 95% CI sMAdCAM-1 <0.41 ng/mL [undetectable] 0.004 6.5 1.3–19.4 0.006 7.5 1.3–25.9 Retinoic acid <1.05 ng/mL 0.045 2.3 1.1–14.5 0.062 2.5 0.9–40.3 Vedolizumab trough level >19 µg/mL 0.061 2.7 1.0–20.4 0.045 2.2 1.1–30.5 Age >38 years 0.025 3.1 1.2–8.7 0.136 2.3 0.8–6.9 Weight >63 kg 0.60 1.4 0.5–3.9 Non-optimized vs optimized therapy 0.083 3.1 0.9–12.7 0.240 1.9 0.5–12.6 Ulcerative colitis vs Crohn’s disease 0.078 6.7 0.9–69.7 0.200 0.5 0.1–1.4 CI: confidence interval; OR: odds ratio; sMAdCAM-1: soluble mucosal addressin cell adhesion molecule-1: TNF: tumor necrosis factor. View Large Pharmacokinetic profiles combining VTL [≤ or >19 µg/mL] and sMAdCAM-1 undetectable or detectable [< or ≥0.41 ng/mL] concentrations are reported in Figure 3, with the proportions of corresponding assays sampled in clinical remission at W30 or at the time of relapse [remission, n = 38; relapse, n = 24]. Clinical remission was reported in 95.2% of assays with VTL > 19 µg/mL and undetectable sMAdCAM-1 [p = 0.005]. Pharmacokinetic profile combining VTL ≤ 19 µg/mL and sMAdCAM-1 ≥ 0.41 ng/mL was significantly associated with a majority of relapses [p = 0.01]. A VTL of ≤ 19 µg/mL and undetectable sMAdCAM-1 [< 0.41 ng/mL] was the most frequent pharmacokinetic profile [~45% of assays performed in clinical remission and 41% at the time of relapse], but it was not discriminant between patients sampled in clinical remission or at the time of relapse. Likewise, VTL > 19 µg/mL and sMAdCAM-1 ≥ 0.41 ng/mL profile was not discriminant. Overall, 79% of assays performed in patients in clinical remission were associated with undetectable sMAdCAM-1 and 74% of assays in patients in relapse were associated with VTL ≤ 19 µg/mL. Figure 3. View largeDownload slide Comparison of pharmacokinetic profiles combining sMAdCAM-1 and vedolizumab trough levels in assays performed in clinical remission or at the time of relapse during vedolizumab maintenance therapy. VTL: vedolizumab trough levels; sMAdCAM-1: soluble mucosal addressin cell adhesion molecule-1. Figure 3. View largeDownload slide Comparison of pharmacokinetic profiles combining sMAdCAM-1 and vedolizumab trough levels in assays performed in clinical remission or at the time of relapse during vedolizumab maintenance therapy. VTL: vedolizumab trough levels; sMAdCAM-1: soluble mucosal addressin cell adhesion molecule-1. The majority of sMAdCAM-1 values were < 0.41 ng/mL [undetectable]. Hence, no quartile analysis was possible for sMAdCAM-1. The comparison of quartiles of VTL in assays performed in clinical remission or at the time of relapse showed a majority of clinical remission across quartiles of VTL, except for VTL ranging from 7.7 to 14.2 µg/mL [p = 0.02 using the one-sided Cochrane–Armitage trend test across all quartiles] [Figure 4], thereby underlying the lack of specificity of VTL for the diagnosis of clinical remission. Regarding RA concentrations, there was no difference in concentration quartiles between assays performed in clinical remission or at the time of relapse [Figure 4]. Figure 4. View largeDownload slide Comparison of quartiles of vedolizumab trough levels and retinoic acid concentrations in assays performed in clinical remission or at the time of relapse during vedolizumab maintenance therapy. The one-sided Cochrane–Armitage trend test was used to compare quartiles. Figure 4. View largeDownload slide Comparison of quartiles of vedolizumab trough levels and retinoic acid concentrations in assays performed in clinical remission or at the time of relapse during vedolizumab maintenance therapy. The one-sided Cochrane–Armitage trend test was used to compare quartiles. 3.2. Longitudinal analysis Median concentrations before VDZ induction were 40.5 ng/mL [23.6–52.8] and 1.7 ng/mL [1.4–2.0] for sMAdCAM-1 and RA, respectively [Figure 5]. The median baseline concentrations of sMAdCAM-1 did not differ between patients in clinical remission during the follow-up and those who relapsed [43.3 vs 32.9 ng/mL, p = 0.33]. In contrast, the median baseline concentrations of RA were significantly different between patients in clinical remission during the follow-up and those who relapsed [1.98 ng/mL [1.67–2.59] vs 1.12 ng/mL [0.94–1.51], p = 0.016]. An optimal baseline RA cut-off of >1.86 ng/mL predicted the clinical remission at W30 with a sensitivity, specificity, accuracy, PPV, and NPV of 73.5%, 88.2%, 77.0%, 91.4%, and 62.0%, respectively (AUROC: 80.7% [59.7–100.0%]). Figure 5. View largeDownload slide Comparison of sMAdCAM-1 and retinoic acid concentrations at Week 0 [before vedolizumab] and during vedolizumab maintenance therapy. The box plots show median, upper and lower quartiles of the data; the whiskers indicate the 95% confidence interval of the values. Figure 5. View largeDownload slide Comparison of sMAdCAM-1 and retinoic acid concentrations at Week 0 [before vedolizumab] and during vedolizumab maintenance therapy. The box plots show median, upper and lower quartiles of the data; the whiskers indicate the 95% confidence interval of the values. Median values of sMAdCAM-1 and RA concentrations were significantly higher [p = 0.0001 using the paired Mann–Whitney test] before VDZ therapy than during the follow-up [sMAdCAM-1: 40.5 vs < 0.41 ng/mL; RA: 1.7 vs 0.97 ng/mL] [Figure 5]. The delta of sMAdCAM-1 concentrations [difference of concentrations before treatment and during the maintenance therapy] was 39.4 ng/mL in patients in clinical remission at W30 compared with 23.9 ng/mL in patients who relapsed [p = 0.003]. The delta of RA concentrations was also higher in patients in clinical remission than in those who relapsed during the follow-up [1.5 ng/mL vs 0.7 ng/mL, p = 0.18]. 4. Discussion For the first time, we showed that undetectable sMAdCAM-1 concentrations during maintenance therapy were strongly associated with clinical remission in IBD patients treated with VDZ. With the longitudinal analysis, we also showed that sMAdCAM-1 was statistically higher before induction of VDZ than during VDZ maintenance therapy, and that its decrease was more pronounced in patients who were in clinical remission during the follow-up. MAdCAM-1 is overexpressed on gut endothelium in active IBD and is upregulated by TNFα.14 Elevated levels of sMAdCAM-1 mirror the higher expression of MAdCAM-1. The decrease of sMAdCAM-1 during VDZ therapy argues for its downregulation under VDZ, as previously demonstrated by blocking TNFα and lymphotoxin-β receptor activation.14,23 We also showed that pharmacokinetics combining undetectable sMAdCAM-1 and VTL > 19 µg/mL were independently associated with clinical remission, and this combination was a favorable pharmacokinetic profile with a PPV of 95%. In a recent Phase II placebo-controlled trial evaluating an anti-MAdCAM-1 antibody in UC patients [TURANDOT study],16 sMAdCAM-1 concentration was measured at baseline and Week 12. A decrease in sMAdCAM-1 concentration was shown in active treatment, but not under placebo, hence supporting our hypothesis and the growing interest in this biomarker in this setting. Moreover, in the longitudinal analysis, high concentrations of RA before VDZ were predictive of VDZ efficacy, probably because of its ability to induce higher expression of α4β7 integrin on lymphocyte surfaces, as previously described in patients with human immunodeficiency virus.24 The situation is similar in patients treated with anti-TNFα in whom the response rates to anti-TNF therapy are higher in patients with high numbers of membrane-bound TNF.25 Indeed, RA is a key physiological factor involved in the induction of α4β7 integrin on lymphocyte surfaces, which is targeted by VDZ.18,26,27 It has also been reported that CD14+ macrophages from the intestinal mucosa of patients with CD are capable of generating RA, which might increase the inflammatory phenotype of these cells.28 During maintenance therapy, a VTL > 19 µg/mL was associated on multivariate analysis with clinical remission, which agreed with previous data from the literature.9,10 However, quartile analyses underlined the lack of specificity of VTL for the diagnosis of clinical remission. Measurement of albumin in blood was not performed in our study, but in contrast to anti-TNF, the impact of serum albumin levels on VTL remains unclear.29,30 The median weight of the included patients was 63 kg [range 50–88], and it was not associated with clinical response on univariate analysis. This could be due to the absence in our cohort of patients with ‘extreme weight’ [>120 kg], since overweight has been identified as a potential clinically important predictor of increased clearance of VDZ.30 Our study has some limitations. The number of included patients was rather low, which precluded meaningful analyses of subgroups, in particular for quartile analyses, evaluation of treatment optimization, and difference between CD and UC. Likewise, the potential diagnostic value of low RA concentrations for clinical remission could not be assessed from our study due to its lack of statistical power. The short duration of maintenance therapy and the retrospective nature of the study could also be pointed out as limits. However, the included patients had a clinical assessment of the disease activity reported prospectively, with objective parameters such as fCal and endoscopic evaluation at the same time decreasing this limitation. Management of patients was based on these clinical scores and independent of the results of VTL, sMAdCAM-1, and RA concentrations. Our work should be considered as a proof-of-concept study, and the analyzed biomarkers can be easily measured in the future with high reproducibility and feasibility in clinical practice. In an ongoing prospective study, we are measuring sMAdCAM-1 and RA concentrations in order to assess their predictive value for clinical remission during VDZ therapy. Awaiting further studies, we can only propose with caution the following strategy. Before starting VDZ therapy, RA concentrations should be measured to help decision-making during the follow-up. Pharmacokinetics combining values of VTL and sMAdCAM-1 should be used if there is a loss of response during maintenance therapy: in the case of a ‘favorable’ pharmacokinetic profile [undetectable sMAdCAM-1 and a VTL of >19 µg/mL], the switch to therapeutic class other than VDZ should be recommended. Optimization of treatment should be proposed if sMAdCAM-1 is detectable with low VTL. With other pharmacokinetic profiles [VTL > 19 µg/mL and detectable sMAdCAM-1; VTL ≤ 19 µg/mL and undetectable sMAdCAM-1], it appears necessary to assess the probability of clinical remission under VDZ: if RA before induction therapy is <1.86 ng/mL, this probability should be low, and a switch of therapeutic class should be proposed. In contrast, if RA is >1.86 ng/mL, optimization of VDZ therapy should be tried. At the moment, this algorithm is to be used with caution, and further studies are needed to improve its level of evidence. In conclusion, undetectable sMAdCAM-1 appears strongly associated with clinical remission during VDZ maintenance therapy in IBD patients. sMAdCAM-1 is actually an inflammatory surrogate marker that probably does not interfere with the effectiveness of VDZ. We observe that it is higher in responder patients before treatment, because the expression of sMAdCAM-1 is also correlated with the expression of MAdCAM-1 on vessels. This indicates that the inflammatory process involved is dependent on lymphocyte migration via this system. This surrogate marker [sMAdCAM-1], however, is unique in predicting the response to VDZ, which is not the case for CRP or fCal. During maintenance therapy, the rate of sMAdCAM-1 falls in responder patients, because it ultimately reflects the decrease in inflammation in these patients. This decreased expression of MAdCAM-1 has already been observed in patients receiving anti-TNF induction therapy.14 Both these induction and maintenance observations suggest that sMAdCAM-1 may bind to α4β7 without inhibiting VDZ uptake. In addition, a pharmacokinetic profile combining undetectable sMAdCAM-1 with high VTL might be potentially be an indicator for TDM. Concentrations of RA before induction are predictive of clinical remission under VDZ therapy. Our findings need to be confirmed in further prospective studies that allow the elaboration of a decisional algorithm. Our data also show that the pharmacokinetics of VDZ are quite complicated and involve different actors/molecules. A mechanistic study of the different molecules involved in the efficacy of VDZ seems to be crucial in order to optimize this treatment in IBD patients. Funding None. Conflict of Interest No conflicts of interest were declared. Author Contributions SP*: conception of the study, data collection, and drafting of the manuscript; NW*: statistical analysis, interpretation of data, and drafting of the manuscript; TDB: data collection; AEB, GB, JF, EDT, BF*: interpretation of data, and critical review of the manuscript; SN: data collection, and critical review of the manuscript; XR*: conception and design of the study, analysis of data, drafting of the manuscript, and study supervision. Abbreviations CD Crohn’s disease CI Confidence interval CRP C-reactive protein DCs Dendritic cells fCal Fecal calprotectin HBI Harvey–Bradshaw index IBD Inflammatory bowel disease NPV Negative predictive value OR Odds ratio PPV Positive predictive value RA Retinoid acid RALDH Retinaldehyde dehydrogenase ROC Receiver operating characteristic Se Sensitivity sMAdCAM-1 Soluble mucosal addressin cell adhesion molecule-1 Sp Specificity TDM Therapeutic drug monitoring TNF Tumor necrosis factor UC Ulcerative colitis VTL Vedolizumab trough level. References 1. Feagan BG , Rutgeerts P , Sands BE , et al. Vedolizumab as induction and maintenance therapy for ulcerative colitis . N Engl J Med 2013 ; 369 : 699 – 710 . Google Scholar CrossRef Search ADS PubMed 2. Sandborn WJ , Feagan BG , Rutgeerts P , et al. Vedolizumab as induction and maintenance therapy for Crohn’s disease . N Engl J Med 2013 ; 369 : 711 – 21 . Google Scholar CrossRef Search ADS PubMed 3. Amiot A , Serrero M , Peyrin-Biroulet L , et al. One-year effectiveness and safety of vedolizumab therapy for inflammatory bowel disease: a prospective multicentre cohort study . Aliment Pharmacol Ther 2017 ; 46 : 310 – 21 . Google Scholar CrossRef Search ADS PubMed 4. Bye WA , Jairath V , Travis SPL . Systematic review: the safety of vedolizumab for the treatment of inflammatory bowel disease . Aliment Pharmacol Ther 2017 ; 46 : 3 – 15 . Google Scholar CrossRef Search ADS PubMed 5. Paul S , Del Tedesco E , Marotte H , et al. Therapeutic drug monitoring of infliximab and mucosal healing in inflammatory bowel disease: a prospective study . Inflamm Bowel Dis 2013 ; 19 : 2568 – 76 . Google Scholar CrossRef Search ADS PubMed 6. Roblin X , Rinaudo M , Del Tedesco E , et al. Development of an algorithm incorporating pharmacokinetics of adalimumab in inflammatory bowel diseases . Am J Gastroenterol 2014 ; 109 : 1250 – 6 . Google Scholar CrossRef Search ADS PubMed 7. Williet N , Paul S , Peyrin-Biroulet L , et al. Pharmacokinetics of infliximab and reduction of treatment for inflammatory bowel diseases . Dig Dis Sci 2016 ; 61 : 990 – 5 . Google Scholar CrossRef Search ADS PubMed 8. Williet N , Boschetti G , Fovet M , et al. Association between low trough levels of vedolizumab during induction therapy for inflammatory bowel diseases and need for additional doses within 6 months . Clin Gastroenterol Hepatol 2017 ; 15 : 1750 – 7 . Google Scholar CrossRef Search ADS PubMed 9. Ungaro RC , Jossen J , Phan BL , et al. Higher vedolizumab trough levels associated with remission in inflammatory bowel disease [IBD] during maintenance therapy . Gastroenterology 2017 ; 152 : S384 – 5 . Google Scholar CrossRef Search ADS 10. Yarur AJ , Bruss A , Jain A , et al. Higher vedolizumab levels are associated with deep remission in patients with Crohn’s disease and ulcerative colitis on maintenance therapy with vedolizumab . J Crohns Colitis 2017 ; 11 : S38 . https://doi.org/10.1093/ecco-jcc/jjx002.057. Accessed February 17, 2017. Google Scholar CrossRef Search ADS 11. Rosario M , French JL , Dirks NL , et al. Exposure–efficacy relationships for vedolizumab induction therapy in patients with ulcerative colitis or Crohn’s disease . J Crohns Colitis 2017 ; 11 : 921 – 9 . Google Scholar CrossRef Search ADS PubMed 12. Boden EK , Shows D , Chiorean MV , et al. Integrin α4β7 expression preceding and saturation during vedolizumab therapy correlate with treatment response in inflammatory bowel disease . Gastroenterology 2017 ; 152 : S39 . Google Scholar CrossRef Search ADS 13. Shigematsu T , Specian RD , Wolf RE , et al. MAdCAM mediates lymphocyte–endothelial cell adhesion in a murine model of chronic colitis . Am J Physiol Gastrointest Liver Physiol 2001 ; 281 : G1309 – 15 . Google Scholar CrossRef Search ADS PubMed 14. Biancheri P , Di Sabatino A , Rovedatti L , et al. Effect of tumor necrosis factor-α blockade on mucosal addressin cell-adhesion molecule-1 in Crohn’s disease . Inflamm Bowel Dis 2013 ; 19 : 259 – 64 . Google Scholar CrossRef Search ADS PubMed 15. Xu Y , Hunt NH , Bao S . The correlation between proinflammatory cytokines, MAdCAM-1 and cellular infiltration in the inflamed colon from TNF-alpha gene knockout mice . Immunol Cell Biol 2007 ; 85 : 633 – 9 . Google Scholar CrossRef Search ADS PubMed 16. Vermeire S , Sandborn WJ , Danese S , et al. Anti-MAdCAM antibody [PF-00547659] for ulcerative colitis [TURANDOT]: a phase 2, randomised, double-blind, placebo-controlled trial . Lancet 2017 ; 390 : 135 – 44 . Google Scholar CrossRef Search ADS PubMed 17. Mwanza-Lisulo M , Kelly P . Potential for use of retinoic acid as an oral vaccine adjuvant . Philos Trans R Soc Lond B Biol Sci 2015 ; 19 : 370 [1671] . 18. Iwata M , Hirakiyama A , Eshima Y , et al. Retinoic acid imprints gut-homing specificity on T cells . Immunity 2004 ; 21 : 527 – 38 . Google Scholar CrossRef Search ADS PubMed 19. Hammerschmidt SI , Friedrichsen M , Boelter J , et al. Retinoic acid induces homing of protective T and B cells to the gut after subcutaneous immunization in mice . J Clin Invest 2011 ; 121 : 3051 – 61 . Google Scholar CrossRef Search ADS PubMed 20. Saurer L , McCullough KC , Summerfield A . In vitro induction of mucosa-type dendritic cells by all-trans retinoic acid . J Immunol 2007 ; 179 : 3504 – 14 . Google Scholar CrossRef Search ADS PubMed 21. Mora JR , Iwata M , Eksteen B , et al. Generation of gut-homing IgA-secreting B cells by intestinal dendritic cells . Science 2006 ; 314 : 1157 – 60 . Google Scholar CrossRef Search ADS PubMed 22. Fransén K , Franzén P , Magnuson A , et al. Polymorphism in the retinoic acid metabolizing enzyme CYP26B1 and the development of Crohn’s disease . PloS One 2013 ; 8 : e72739 . Google Scholar CrossRef Search ADS PubMed 23. Stopfer P , Obermeier F , Dunger N , et al. Blocking lymphotoxin-beta receptor activation diminishes inflammation via reduced mucosal addressin cell adhesion molecule-1 [MAdCAM-1] expression and leucocyte margination in chronic DSS-induced colitis . Clin Exp Immunol 2004 ; 136 : 21 – 9 . Google Scholar CrossRef Search ADS PubMed 24. Wacleche VS , Chomont N , Gosselin A , et al. The colocalization potential of HIV-specific CD8+ and CD4+ T-cells is mediated by integrin β7 but not CCR6 and regulated by retinoic acid . PloS One 2012 ; 7 : e32964 . Google Scholar CrossRef Search ADS PubMed 25. Atreya R , Neumann H , Neufert C , et al. In vivo imaging using fluorescent antibodies to tumor necrosis factor predicts therapeutic response in Crohn’s disease . Nat Med 2014 ; 20 : 313 – 8 . Google Scholar CrossRef Search ADS PubMed 26. Eksteen B , Mora JR , Haughton EL , et al. Gut homing receptors on CD8 T cells are retinoic acid dependent and not maintained by liver dendritic or stellate cells . Gastroenterology 2009 ; 137 : 320 – 9 . Google Scholar CrossRef Search ADS PubMed 27. Mora JR , Bono MR , Manjunath N , et al. Selective imprinting of gut-homing T cells by Peyer’s patch dendritic cells . Nature 2003 ; 424 : 88 – 93 . Google Scholar CrossRef Search ADS PubMed 28. Sanders TJ , McCarthy NE , Giles EM , et al. Increased production of retinoic acid by intestinal macrophages contributes to their inflammatory phenotype in patients with Crohn’s disease . Gastroenterology 2014 ; 146 : 1278 – 88 . Google Scholar CrossRef Search ADS PubMed 29. Shivashankar R , Aberra F , Mendoza Ladd AH , et al. Effect of serum albumin levels on efficacy of vedolizumab in patients with Crohn’s disease . Gastroenterology 2017 ; 152[ Suppl. 1]: s407 – 8 . Google Scholar CrossRef Search ADS 30. Rosario M , Dirks NL , Gastonguay MR , et al. Population pharmacokinetics–pharmacodynamics of vedolizumab in patients with ulcerative colitis and Crohn’s disease . Aliment Pharmacol Ther 2015 ; 42 : 188 – 202 . Google Scholar CrossRef Search ADS PubMed Copyright © 2018 European Crohn’s and Colitis Organisation (ECCO). Published by Oxford University Press. All rights reserved. For permissions, please email: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Journal of Crohn's and Colitis Oxford University Press

Soluble Mucosal Addressin Cell Adhesion Molecule 1 and Retinoic Acid are Potential Tools for Therapeutic Drug Monitoring in Patients with Inflammatory Bowel Disease Treated with Vedolizumab: A Proof of Concept Study

Loading next page...
 
/lp/ou_press/soluble-mucosal-addressin-cell-adhesion-molecule-1-and-retinoic-acid-Qnxe9CNF78
Publisher
Oxford University Press
Copyright
Copyright © 2018 European Crohn’s and Colitis Organisation (ECCO). Published by Oxford University Press. All rights reserved. For permissions, please email: journals.permissions@oup.com
ISSN
1873-9946
eISSN
1876-4479
D.O.I.
10.1093/ecco-jcc/jjy077
Publisher site
See Article on Publisher Site

Abstract

Abstract Background and Aims Vedolizumab [VDZ], a humanized monoclonal antibody targeting α4β7 integrin, is effective in induction and maintenance therapy in patients with inflammatory bowel disease [IBD] who have not adequately responded to standard therapies, and high vedolizumab trough levels [VTLs] have been associated with clinical remission. The α4β7 integrin binds to endothelial MAdCAM-1 and is upregulated by retinoic acid [RA]. The aim of this study was to determine the relationships between soluble MAdCAM-1 [sMAdCAM-1] and RA concentrations during clinical remission with VDZ maintenance therapy. Methods In a retrospective study performed in IBD patients treated with VDZ, we measured VTL, sMAdCAM-1 and RA concentrations. Results Among the 62 included patients [38 Crohn’s disease], 24 relapsed and 38 stayed in remission from Weeks 10 to 30 after VDZ initiation. During this maintenance therapy, the median values of VTLs and RA were 15.4 µg/mL and 0.97 ng/mL, respectively, whereas sMAdCAM-1 was undetectable [<0.41 ng/mL] in 67.3% of samples. The positive predictive value [PPV] of undetectable sMAdCAM-1 for clinical remission was 80.0%, with a corresponding sensitivity of 74.6%. On multivariate analysis, undetectable sMAdCAM-1 and high VTLs [>19 µg/mL] were independently associated with clinical remission [OR = 7.5, p = 0.006 and OR = 2.2, p = 0.045, respectively]. The combination of sMAdCAM-1 < 0.41 ng/mL and VTL > 19 µg/mL was the best pharmacokinetic profile, with a PPV of 95.2%. Median values of sMAdCAM-1 and RA were significantly higher [p = 0.0001] before VDZ therapy than during the follow-up [sMAdCAM-1: 40.5 vs < 0.41 ng/mL; RA: 1.7 vs 0.97 ng/mL]. Only RA > 1.86 ng/mL before VDZ therapy was predictive of clinical remission during the follow-up (Area Under a Receiver Operating Characteristic curve [AUROC] = 80.7%). Conclusions Undetectable sMAdCAM-1 appears strongly associated with clinical remission during VDZ maintenance therapy. Combination of undetectable sMAdCAM-1 with high VTL is also potentially interesting for therapeutic drug monitoring. Baseline RA concentrations are predictive of clinical remission. These findings need to be confirmed in further prospective studies. Soluble MAdCAM-1, retinoic acid, vedolizumab, inflammatory bowel disease, therapeutic drug monitoring 1. Introduction In recent decades, biologic therapies such as anti-tumor necrosis factor [TNF] and vedolizumab [VDZ] have been developed for treating patients with inflammatory bowel disease [IBD].1–4 To optimize their use, therapeutic drug monitoring [TDM] has emerged as an interesting approach. Performances of TDM have been so far assessed with anti-TNF agents, and several decision trees based on trough levels of anti-TNF and anti-drug antibodies have been proposed.5–7 VDZ is a humanized monoclonal antibody targeting α4β7 integrin. It modulates inflammation of the gastrointestinal wall by limiting lymphocyte homing to the lamina propria without inducing systemic immunosuppression. Vedolizumab was approved in 2014 for induction and maintenance therapy in IBD patients who had not adequately responded to one or more standard therapies [corticosteroids, immunomodulators, or anti-TNF]. We recently reported the association between low VDZ trough levels [VTLs] during induction therapy and the need for additional doses within 6 months.8 Other recent studies have reported the association of high VTLs with clinical remission.9–12 However, other factors may be involved in the clearance of the drug and in clinical outcomes. To the best of our knowledge, none of these factors has been studied so far. Some authors evaluated the integrin α4β7 expression on multiple lymphocyte subsets before and during VDZ therapy, but the method used [flow cytometry] appears hardly applicable in clinical practice.12 Leukocyte recruitment is pivotal for the initiation and perpetuation of IBD and is controlled by the interactions of chemokines and adhesion molecules. Mucosal addressin cell adhesion molecule 1 [MAdCAM-1] is a cell-surface immunoglobulin superfamily member that promotes the adhesion of T and B cells to vascular endothelium and is critical for lymphocyte homing to the gut.13 Its expression is increased in patients with IBD and in animal colitis models.14,15 In a recent placebo-controlled trial evaluating an anti-MAdCAM-1 antibody, Vermeire et al. reported a correlation between the circulating soluble form of MAdCAM-1 [sMAdCAM-1] and its tissue expression.16 Vitamin A is the term given to a collection of related molecules known as retinoic acid [RA].17 During vitamin A metabolism, the irreversible conversion of retinal to RA is catalyzed by retinaldehyde dehydrogenase [RALDH], which is expressed by dendritic cells [DCs] from Peyer patches and mesenteric lymph nodes.18 The RALDH allows intestinal DCs to convert retinal to RA, which in turn induces the expression of the gut homing receptor α4β7 on T cells.18–20 Upregulation of α4β7 by RA on the surface of B cells has also been demonstrated.21 Polymorphism in the CYP26B1 enzyme is likely involved in the degradation of RA and has been described in patients with Crohn’s disease [CD], supporting the role of vitamin A in the pathophysiology of CD.22 In this study, we retrospectively examined the relationships between sMAdCAM-1 and RA concentrations during the response to VDZ in IBD patients. We hypothesized that patients with high levels of circulating RA could be potential good responders to VDZ, because RA induces the lymphocyte expression of α4β7, which is the target of VDZ. In contrast, in the presence of high levels of sMAdCAM-1, the α4β7 integrin expressed at the lymphocyte surface could be overloaded with sMAdCAM-1, reducing the accessibility of the integrin to VDZ, and this could therefore be associated with a lack of response to VDZ. 2. Methods 2.1. Patients All consecutive IBD patients from three referral centers who were treated with VDZ as maintenance therapy and in clinical remission after induction therapy were retrospectively included in the present study. They had failed or had developed intolerance to at least one line of anti-TNF. Three hundred milligrams of VDZ were administered at Weeks [W] 0, 2, and 6 as induction therapy, and then were administered once every 8 weeks during the maintenance therapy. Serum samples during the induction phase were not collected. At W10, clinical activity was assessed, and patients in whom clinical remission was not achieved received an additional dose of 300 mg VDZ at W10 followed by an infusion of VDZ once every 4 weeks as optimization of the treatment during the maintenance therapy. Clinical activity was assessed at W10 in all patients, at W30 in patients in remission, at W14 in optimized patients and in patients who relapsed before W30 (using the Harvey–Bradshaw index [HBI] and faecal calprotectin [fCal] in patients with CD, and the Mayo score in patients with ulcerative colitis [UC]). Clinical remission was defined as a HBI of ≤3 with fCal levels of <250 µg/g stools for CD, and as a Mayo score of ≤3 with an endoscopic subscore of 0 to 1 for UC without concomitant corticosteroids. Primary non-response to VDZ at W14 and relapse were defined by a HBI of >4 with fCal of >250 µg/g stools for CD, and by a Mayo score of >4 with an endoscopic subscore of >1 for UC. Patients with primary non-response to VDZ at W14 in spite of VDZ infusion at W10 were excluded from our study, which focused on maintenance therapy. A medical visit was performed before each VDZ infusion. In all patients, sMAdCAM-1 and RA concentrations were measured in the blood before the first VDZ infusion. In patients in clinical remission, measurements of VTL, sMAdCAM-1, and RA concentrations were performed at W30. For patients in clinical relapse after W10, measurements of VTL, sMAdCAM-1, and RA concentrations were performed at the time of relapse. Measurements of VDZ, sMAdCAM-1, and RA concentrations were performed retrospectively. Hence, the management of patients was independent of these results. All patients signed an informed written consent to the protocol, which was approved by the Ethics Committee of Saint-Etienne University and Centre National Informatique et Liberté [CNIL 1849323 v 0]. 2.2. Measurements of vedolizumab, soluble MAdCAM-1, and retinoic acid concentrations Human sMAdCAM-1 concentrations were measured using the HycultBiotech ELISA assay [Clinisciences, Montrouge, France]. Briefly, the human sMAdCAM-1 ELISA is a ready to use solid-phase sandwich ELISA that allows the determination of sMAdCAM-1 concentrations between 0.41 and 100 ng/mL. We tested the spike of VDZ at a range of between 1 and 30 µg/ml in a V0 sample with a high concentration of sMAdCAM-1 [<5%]. We did not observe any significant differences in sMAdCAM-1 with or without VDZ. Moreover, the situation was exactly the same when dosing withVDZ in the presence of sMAdCAM-1. Again, we did not observe any interference in VDZ measurement in the presence of sMAdCAM-1, as in the V0 samples. Human RA concentrations were measured with the CUSABIO competitive inhibition ELISA assay [Clinisciences, Montrouge, France]. The detection range of this assay is from 0.42 to 10 ng/mL. Vedolizumab and anti-VDZ antibodies were measured using the Lisa-Tracker duo VDZ ELISA assay [Theradiag, Marne-la-Vallée, France]. The assay ranges from 2 to 60 µg/mL for VDZ and from 35 to 500 ng/mL for anti-VDZ antibodies. We used in this study a drug-sensitive ELISA assay for monitoring anti-VDZ antibodies. This assay has been designed to reduce the formation of complexes between VDZ and anti-VDZ antibodies by using specific buffers. However, with the drug-sensitive ELISA technique, when VDZ is detectable in the serum, the value of the anti-VDZ antibody is considered ‘inconclusive’. 2.3. Statistical analysis The primary objective was to evaluate the sMAdCAM-1 and RA concentration, in addition to the VTL, and to assess clinical remission under VDZ maintenance therapy. First, a cross-sectional analysis was carried out including only measurements of sMAdCAM-1, RA, and VTL [performed during maintenance therapy at W30 or at the time of relapse]. Linear correlations between VDZ, sMAdCAM-1, or RA concentrations were assessed using the Pearson correlation coefficient. The most associated threshold with clinical remission for each biological variable were identified by receiver operating characteristic [ROC] curves. Factors associated with clinical remission were then identified using uni- and multivariate logistic regressions, including dichotomized variables defined by previously identified thresholds. Variables that achieved a p < 0.1 value on univariate analysis were included in the multivariate analysis. Pharmacokinetic profiles combining dichotomized values significantly associated with clinical remission on multivariate analysis were then tested according to proportions of patients in remission or relapse, and were compared with the Chi2 or Fisher tests, as appropriate. Similar analyses focusing on quartiles of the biomarkers were also performed. The one-sided Cochrane–Armitage trend test was used to compare quartiles. Second, in a longitudinal analysis, the median concentrations of sMAdCAM-1 and RA before and during VDZ maintenance therapy were compared using the paired Mann–Whitney test. Predictive performances for clinical remission of these biomarkers (sensitivity, specificity, accuracy, positive predictive value [PPV], and negative predictive value [NPV]) were tested by ROC curves. Statistical analysis was performed using IBM SPSS 20.0.0 [IBM, Somers, NY]. The significance level was defined as p ≤ 0.05. 3. Results 3.1. Cross-sectional analysis Characteristics of the population study are reported in Table 1. Among the 62 patients in clinical remission at W14 [38 CD] considered for analysis, ~50% were female, and the median age was 38 years [median; IQR25–75: 30–56]. All were previously treated with immunosuppressive drugs [azathioprine or methotrexate] and usually two lines of anti-TNF agents. The median interval between anti-TNF discontinuation and VDZ initiation was 4 weeks [IQR25–75: 4–6] and 8 weeks [IQR25–75: 8–12] for adalimumab and infliximab, respectively. Residual trough levels of anti-TNF were probably low in this study and we did not expect a strong carryover effect. From W10 to W30, 24 patients relapsed, whereas 38 stayed in clinical remission [Figure 1]. Sampling was performed at the time of relapse in patients who relapsed and was therefore not at trough time in all the 24 patients. Sampling was not performed during the induction phase. However, in these cases, sampling was performed at least 4 weeks after the last infusion. Conversely, sampling was performed 8 weeks after the last infusion in 28 patients in clinical remission with no optimization, and 4 weeks after the last infusion in 12 patients optimized and in clinical remission. A total of 124 assays were performed, with 62 during VDZ maintenance therapy and 62 before the first infusion of VDZ. During maintenance therapy, 38 determinations of VTL, sMAdCAM-1, and RA concentrations were performed from patients in clinical remission and 24 from patients who relapsed [Figure 1]. Table 1. Characteristics of the population. Population study [n = 62] Women, n [%] 32 [51.6] Age year, median [IQR25–75] 38 [30–56] Weight kg, median [IQR25–75] 63 [58–74] Crohn,n [%] 38 [61.3] Age at diagnosis  <17 years [A1] 6 [15.8]  17–40 years [A2] 24 [63.2]  above 40 years [A3] 8 [21.0] Behaviour*, n [%]  non-stricturing, non-penetrating [B1] 17 [44.7]  stricturing [B2] 17 [44.7]  penetrating [B3] 4 [10.6] Location*, n [%]  ileal [L1] 13 [34.2]  colonic [L2] 4 [10.5]  ileocolonic [L3] 21 [55.3]  perianal disease 7 [18.4] Ulcerative colitis, n [%]* 24 [38.7]  proctitis [E1] 2 [8.3]  left sided [E2] 4 [16.7]  extensive [E3] 18 [75.0] Previous medication, n [%] Azathioprine 48 [77.4] Methotrexate 14 [22.6] Anti-TNF 62 [100]  1 line anti-TNF 11 [17.7]  2 lines anti-TNF 51 [82.3] History of surgery for IBD,n [%] 27 [43.5] Population study [n = 62] Women, n [%] 32 [51.6] Age year, median [IQR25–75] 38 [30–56] Weight kg, median [IQR25–75] 63 [58–74] Crohn,n [%] 38 [61.3] Age at diagnosis  <17 years [A1] 6 [15.8]  17–40 years [A2] 24 [63.2]  above 40 years [A3] 8 [21.0] Behaviour*, n [%]  non-stricturing, non-penetrating [B1] 17 [44.7]  stricturing [B2] 17 [44.7]  penetrating [B3] 4 [10.6] Location*, n [%]  ileal [L1] 13 [34.2]  colonic [L2] 4 [10.5]  ileocolonic [L3] 21 [55.3]  perianal disease 7 [18.4] Ulcerative colitis, n [%]* 24 [38.7]  proctitis [E1] 2 [8.3]  left sided [E2] 4 [16.7]  extensive [E3] 18 [75.0] Previous medication, n [%] Azathioprine 48 [77.4] Methotrexate 14 [22.6] Anti-TNF 62 [100]  1 line anti-TNF 11 [17.7]  2 lines anti-TNF 51 [82.3] History of surgery for IBD,n [%] 27 [43.5] IQR: interquartile range; TNF: tumor necrosis factor; *according to the Montreal Classification. View Large Table 1. Characteristics of the population. Population study [n = 62] Women, n [%] 32 [51.6] Age year, median [IQR25–75] 38 [30–56] Weight kg, median [IQR25–75] 63 [58–74] Crohn,n [%] 38 [61.3] Age at diagnosis  <17 years [A1] 6 [15.8]  17–40 years [A2] 24 [63.2]  above 40 years [A3] 8 [21.0] Behaviour*, n [%]  non-stricturing, non-penetrating [B1] 17 [44.7]  stricturing [B2] 17 [44.7]  penetrating [B3] 4 [10.6] Location*, n [%]  ileal [L1] 13 [34.2]  colonic [L2] 4 [10.5]  ileocolonic [L3] 21 [55.3]  perianal disease 7 [18.4] Ulcerative colitis, n [%]* 24 [38.7]  proctitis [E1] 2 [8.3]  left sided [E2] 4 [16.7]  extensive [E3] 18 [75.0] Previous medication, n [%] Azathioprine 48 [77.4] Methotrexate 14 [22.6] Anti-TNF 62 [100]  1 line anti-TNF 11 [17.7]  2 lines anti-TNF 51 [82.3] History of surgery for IBD,n [%] 27 [43.5] Population study [n = 62] Women, n [%] 32 [51.6] Age year, median [IQR25–75] 38 [30–56] Weight kg, median [IQR25–75] 63 [58–74] Crohn,n [%] 38 [61.3] Age at diagnosis  <17 years [A1] 6 [15.8]  17–40 years [A2] 24 [63.2]  above 40 years [A3] 8 [21.0] Behaviour*, n [%]  non-stricturing, non-penetrating [B1] 17 [44.7]  stricturing [B2] 17 [44.7]  penetrating [B3] 4 [10.6] Location*, n [%]  ileal [L1] 13 [34.2]  colonic [L2] 4 [10.5]  ileocolonic [L3] 21 [55.3]  perianal disease 7 [18.4] Ulcerative colitis, n [%]* 24 [38.7]  proctitis [E1] 2 [8.3]  left sided [E2] 4 [16.7]  extensive [E3] 18 [75.0] Previous medication, n [%] Azathioprine 48 [77.4] Methotrexate 14 [22.6] Anti-TNF 62 [100]  1 line anti-TNF 11 [17.7]  2 lines anti-TNF 51 [82.3] History of surgery for IBD,n [%] 27 [43.5] IQR: interquartile range; TNF: tumor necrosis factor; *according to the Montreal Classification. View Large Figure 1. View largeDownload slide Description of the study design. VTLs: vedolizumab trough levels; sMAdCAM-1: soluble mucosal addressin cell adhesion molecule 1; CD: Crohn’s disease. Figure 1. View largeDownload slide Description of the study design. VTLs: vedolizumab trough levels; sMAdCAM-1: soluble mucosal addressin cell adhesion molecule 1; CD: Crohn’s disease. During maintenance therapy, the median VTL was 15.4 µg/mL [IQR25–75: 7.7–26.5] and there was no anti-VDZ antibody. Median sMAdCAM-1 was significantly lower [p < 0.01] in patients who stayed in remission compared with in those who relapsed [Figure 2]. Soluble MAdCAM-1 was undetectable [< 0.41 ng/mL] in 67.3% of samples [IQR25–75: 0.00–1.29], and the highest level of sMAdCAM-1 was 16.54 ng/mL; it was undetectable in 87.4% of patients in clinical remission compared with in 45.1% of relapsing patients [p = 0.003]. The median RA concentration was 0.97 ng/mL [IQR25–75: 0.42–1.10], and RA was undetectable [< 0.42 ng/mL] in 25.2% of samples. There was no linear correlation between these biomarkers [Pearson’s correlation coefficient: 0.20 and – 0.07 between VTL and sMAdCAM-1 and between VTL and RA, respectively; –0.10 between sMAdCAM-1 and RA]. Figure 2. View largeDownload slide sMAdCAM-1 [A], retinoic acid [B] and vedolizumab trough level [C] concentrations in patients who relapsed or who stayed in remission during vedolizumab maintenance therapy. sMAdCAM-1: soluble mucosal addressin cell adhesion molecule 1. The box plots show median, upper, and lower quartiles of the data; the whiskers indicate the 95% confidence interval of the values. Median and interquartile ranges [25–75] are indicated below the graphs. Figure 2. View largeDownload slide sMAdCAM-1 [A], retinoic acid [B] and vedolizumab trough level [C] concentrations in patients who relapsed or who stayed in remission during vedolizumab maintenance therapy. sMAdCAM-1: soluble mucosal addressin cell adhesion molecule 1. The box plots show median, upper, and lower quartiles of the data; the whiskers indicate the 95% confidence interval of the values. Median and interquartile ranges [25–75] are indicated below the graphs. The optimal thresholds of biomarkers associated with clinical remission with their diagnostic performances were determined by ROC curves and reported in Table 2. The most accurate biomarker was undetectable sMAdCAM-1 [threshold < 0.41 ng/mL]. On univariate analysis by logistic regression [Table 3], undetectable sMAdCAM-1, age > 38 years, RA < 1.05 ng/mL, VTL > 19 µg/mL, type of IBD [UC vs CD], and optimization of the treatment were associated with clinical remission during the follow-up and were included on the multivariate analysis. On multivariate analysis, only undetectable sMAdCAM-1 and VTL > 19 µg/mL remained independently associated with clinical remission [OR = 7.5, 95% CI: 1.3–25.9, p = 0.006 and OR = 2.2, 95% CI: 1.1–30.5, p = 0.045, respectively], with a trend toward a significant association for RA < 1.05 ng/mL [p = 0.062]. Table 2. Identification of optimal thresholds associated with clinical remission for vedolizumab trough levels, sMAdCAM-1 and retinoic acid by receiver operating characteristic [ROC] curves during vedolizumab maintenance therapy. Threshold Sensitivity Specificity Accuracy PPV NPV Vedolizumab trough level >19.0 µg/mL 47.5 65.0 0.50 72.1 39.7 sMAdCAM-1 <0.41 ng/mL [undetectable] 74.6 54.2 0.69 80.0 46.4 Retinoic acid <1.05 ng/mL 59.0 65.6 0.57 78.3 52.0 Threshold Sensitivity Specificity Accuracy PPV NPV Vedolizumab trough level >19.0 µg/mL 47.5 65.0 0.50 72.1 39.7 sMAdCAM-1 <0.41 ng/mL [undetectable] 74.6 54.2 0.69 80.0 46.4 Retinoic acid <1.05 ng/mL 59.0 65.6 0.57 78.3 52.0 sMAdCAM-1: soluble mucosal addressin cell adhesion molecule-1; NPV: negative predictive value; PPV: positive predictive value. View Large Table 2. Identification of optimal thresholds associated with clinical remission for vedolizumab trough levels, sMAdCAM-1 and retinoic acid by receiver operating characteristic [ROC] curves during vedolizumab maintenance therapy. Threshold Sensitivity Specificity Accuracy PPV NPV Vedolizumab trough level >19.0 µg/mL 47.5 65.0 0.50 72.1 39.7 sMAdCAM-1 <0.41 ng/mL [undetectable] 74.6 54.2 0.69 80.0 46.4 Retinoic acid <1.05 ng/mL 59.0 65.6 0.57 78.3 52.0 Threshold Sensitivity Specificity Accuracy PPV NPV Vedolizumab trough level >19.0 µg/mL 47.5 65.0 0.50 72.1 39.7 sMAdCAM-1 <0.41 ng/mL [undetectable] 74.6 54.2 0.69 80.0 46.4 Retinoic acid <1.05 ng/mL 59.0 65.6 0.57 78.3 52.0 sMAdCAM-1: soluble mucosal addressin cell adhesion molecule-1; NPV: negative predictive value; PPV: positive predictive value. View Large Table 3. Uni- and multivariate analysis by logistic regression to identify associated factors with clinical remission during vedolizumab maintenance therapy. Univariate analysis Multivariate analysis p value OR 95% CI p value OR 95% CI sMAdCAM-1 <0.41 ng/mL [undetectable] 0.004 6.5 1.3–19.4 0.006 7.5 1.3–25.9 Retinoic acid <1.05 ng/mL 0.045 2.3 1.1–14.5 0.062 2.5 0.9–40.3 Vedolizumab trough level >19 µg/mL 0.061 2.7 1.0–20.4 0.045 2.2 1.1–30.5 Age >38 years 0.025 3.1 1.2–8.7 0.136 2.3 0.8–6.9 Weight >63 kg 0.60 1.4 0.5–3.9 Non-optimized vs optimized therapy 0.083 3.1 0.9–12.7 0.240 1.9 0.5–12.6 Ulcerative colitis vs Crohn’s disease 0.078 6.7 0.9–69.7 0.200 0.5 0.1–1.4 Univariate analysis Multivariate analysis p value OR 95% CI p value OR 95% CI sMAdCAM-1 <0.41 ng/mL [undetectable] 0.004 6.5 1.3–19.4 0.006 7.5 1.3–25.9 Retinoic acid <1.05 ng/mL 0.045 2.3 1.1–14.5 0.062 2.5 0.9–40.3 Vedolizumab trough level >19 µg/mL 0.061 2.7 1.0–20.4 0.045 2.2 1.1–30.5 Age >38 years 0.025 3.1 1.2–8.7 0.136 2.3 0.8–6.9 Weight >63 kg 0.60 1.4 0.5–3.9 Non-optimized vs optimized therapy 0.083 3.1 0.9–12.7 0.240 1.9 0.5–12.6 Ulcerative colitis vs Crohn’s disease 0.078 6.7 0.9–69.7 0.200 0.5 0.1–1.4 CI: confidence interval; OR: odds ratio; sMAdCAM-1: soluble mucosal addressin cell adhesion molecule-1: TNF: tumor necrosis factor. View Large Table 3. Uni- and multivariate analysis by logistic regression to identify associated factors with clinical remission during vedolizumab maintenance therapy. Univariate analysis Multivariate analysis p value OR 95% CI p value OR 95% CI sMAdCAM-1 <0.41 ng/mL [undetectable] 0.004 6.5 1.3–19.4 0.006 7.5 1.3–25.9 Retinoic acid <1.05 ng/mL 0.045 2.3 1.1–14.5 0.062 2.5 0.9–40.3 Vedolizumab trough level >19 µg/mL 0.061 2.7 1.0–20.4 0.045 2.2 1.1–30.5 Age >38 years 0.025 3.1 1.2–8.7 0.136 2.3 0.8–6.9 Weight >63 kg 0.60 1.4 0.5–3.9 Non-optimized vs optimized therapy 0.083 3.1 0.9–12.7 0.240 1.9 0.5–12.6 Ulcerative colitis vs Crohn’s disease 0.078 6.7 0.9–69.7 0.200 0.5 0.1–1.4 Univariate analysis Multivariate analysis p value OR 95% CI p value OR 95% CI sMAdCAM-1 <0.41 ng/mL [undetectable] 0.004 6.5 1.3–19.4 0.006 7.5 1.3–25.9 Retinoic acid <1.05 ng/mL 0.045 2.3 1.1–14.5 0.062 2.5 0.9–40.3 Vedolizumab trough level >19 µg/mL 0.061 2.7 1.0–20.4 0.045 2.2 1.1–30.5 Age >38 years 0.025 3.1 1.2–8.7 0.136 2.3 0.8–6.9 Weight >63 kg 0.60 1.4 0.5–3.9 Non-optimized vs optimized therapy 0.083 3.1 0.9–12.7 0.240 1.9 0.5–12.6 Ulcerative colitis vs Crohn’s disease 0.078 6.7 0.9–69.7 0.200 0.5 0.1–1.4 CI: confidence interval; OR: odds ratio; sMAdCAM-1: soluble mucosal addressin cell adhesion molecule-1: TNF: tumor necrosis factor. View Large Pharmacokinetic profiles combining VTL [≤ or >19 µg/mL] and sMAdCAM-1 undetectable or detectable [< or ≥0.41 ng/mL] concentrations are reported in Figure 3, with the proportions of corresponding assays sampled in clinical remission at W30 or at the time of relapse [remission, n = 38; relapse, n = 24]. Clinical remission was reported in 95.2% of assays with VTL > 19 µg/mL and undetectable sMAdCAM-1 [p = 0.005]. Pharmacokinetic profile combining VTL ≤ 19 µg/mL and sMAdCAM-1 ≥ 0.41 ng/mL was significantly associated with a majority of relapses [p = 0.01]. A VTL of ≤ 19 µg/mL and undetectable sMAdCAM-1 [< 0.41 ng/mL] was the most frequent pharmacokinetic profile [~45% of assays performed in clinical remission and 41% at the time of relapse], but it was not discriminant between patients sampled in clinical remission or at the time of relapse. Likewise, VTL > 19 µg/mL and sMAdCAM-1 ≥ 0.41 ng/mL profile was not discriminant. Overall, 79% of assays performed in patients in clinical remission were associated with undetectable sMAdCAM-1 and 74% of assays in patients in relapse were associated with VTL ≤ 19 µg/mL. Figure 3. View largeDownload slide Comparison of pharmacokinetic profiles combining sMAdCAM-1 and vedolizumab trough levels in assays performed in clinical remission or at the time of relapse during vedolizumab maintenance therapy. VTL: vedolizumab trough levels; sMAdCAM-1: soluble mucosal addressin cell adhesion molecule-1. Figure 3. View largeDownload slide Comparison of pharmacokinetic profiles combining sMAdCAM-1 and vedolizumab trough levels in assays performed in clinical remission or at the time of relapse during vedolizumab maintenance therapy. VTL: vedolizumab trough levels; sMAdCAM-1: soluble mucosal addressin cell adhesion molecule-1. The majority of sMAdCAM-1 values were < 0.41 ng/mL [undetectable]. Hence, no quartile analysis was possible for sMAdCAM-1. The comparison of quartiles of VTL in assays performed in clinical remission or at the time of relapse showed a majority of clinical remission across quartiles of VTL, except for VTL ranging from 7.7 to 14.2 µg/mL [p = 0.02 using the one-sided Cochrane–Armitage trend test across all quartiles] [Figure 4], thereby underlying the lack of specificity of VTL for the diagnosis of clinical remission. Regarding RA concentrations, there was no difference in concentration quartiles between assays performed in clinical remission or at the time of relapse [Figure 4]. Figure 4. View largeDownload slide Comparison of quartiles of vedolizumab trough levels and retinoic acid concentrations in assays performed in clinical remission or at the time of relapse during vedolizumab maintenance therapy. The one-sided Cochrane–Armitage trend test was used to compare quartiles. Figure 4. View largeDownload slide Comparison of quartiles of vedolizumab trough levels and retinoic acid concentrations in assays performed in clinical remission or at the time of relapse during vedolizumab maintenance therapy. The one-sided Cochrane–Armitage trend test was used to compare quartiles. 3.2. Longitudinal analysis Median concentrations before VDZ induction were 40.5 ng/mL [23.6–52.8] and 1.7 ng/mL [1.4–2.0] for sMAdCAM-1 and RA, respectively [Figure 5]. The median baseline concentrations of sMAdCAM-1 did not differ between patients in clinical remission during the follow-up and those who relapsed [43.3 vs 32.9 ng/mL, p = 0.33]. In contrast, the median baseline concentrations of RA were significantly different between patients in clinical remission during the follow-up and those who relapsed [1.98 ng/mL [1.67–2.59] vs 1.12 ng/mL [0.94–1.51], p = 0.016]. An optimal baseline RA cut-off of >1.86 ng/mL predicted the clinical remission at W30 with a sensitivity, specificity, accuracy, PPV, and NPV of 73.5%, 88.2%, 77.0%, 91.4%, and 62.0%, respectively (AUROC: 80.7% [59.7–100.0%]). Figure 5. View largeDownload slide Comparison of sMAdCAM-1 and retinoic acid concentrations at Week 0 [before vedolizumab] and during vedolizumab maintenance therapy. The box plots show median, upper and lower quartiles of the data; the whiskers indicate the 95% confidence interval of the values. Figure 5. View largeDownload slide Comparison of sMAdCAM-1 and retinoic acid concentrations at Week 0 [before vedolizumab] and during vedolizumab maintenance therapy. The box plots show median, upper and lower quartiles of the data; the whiskers indicate the 95% confidence interval of the values. Median values of sMAdCAM-1 and RA concentrations were significantly higher [p = 0.0001 using the paired Mann–Whitney test] before VDZ therapy than during the follow-up [sMAdCAM-1: 40.5 vs < 0.41 ng/mL; RA: 1.7 vs 0.97 ng/mL] [Figure 5]. The delta of sMAdCAM-1 concentrations [difference of concentrations before treatment and during the maintenance therapy] was 39.4 ng/mL in patients in clinical remission at W30 compared with 23.9 ng/mL in patients who relapsed [p = 0.003]. The delta of RA concentrations was also higher in patients in clinical remission than in those who relapsed during the follow-up [1.5 ng/mL vs 0.7 ng/mL, p = 0.18]. 4. Discussion For the first time, we showed that undetectable sMAdCAM-1 concentrations during maintenance therapy were strongly associated with clinical remission in IBD patients treated with VDZ. With the longitudinal analysis, we also showed that sMAdCAM-1 was statistically higher before induction of VDZ than during VDZ maintenance therapy, and that its decrease was more pronounced in patients who were in clinical remission during the follow-up. MAdCAM-1 is overexpressed on gut endothelium in active IBD and is upregulated by TNFα.14 Elevated levels of sMAdCAM-1 mirror the higher expression of MAdCAM-1. The decrease of sMAdCAM-1 during VDZ therapy argues for its downregulation under VDZ, as previously demonstrated by blocking TNFα and lymphotoxin-β receptor activation.14,23 We also showed that pharmacokinetics combining undetectable sMAdCAM-1 and VTL > 19 µg/mL were independently associated with clinical remission, and this combination was a favorable pharmacokinetic profile with a PPV of 95%. In a recent Phase II placebo-controlled trial evaluating an anti-MAdCAM-1 antibody in UC patients [TURANDOT study],16 sMAdCAM-1 concentration was measured at baseline and Week 12. A decrease in sMAdCAM-1 concentration was shown in active treatment, but not under placebo, hence supporting our hypothesis and the growing interest in this biomarker in this setting. Moreover, in the longitudinal analysis, high concentrations of RA before VDZ were predictive of VDZ efficacy, probably because of its ability to induce higher expression of α4β7 integrin on lymphocyte surfaces, as previously described in patients with human immunodeficiency virus.24 The situation is similar in patients treated with anti-TNFα in whom the response rates to anti-TNF therapy are higher in patients with high numbers of membrane-bound TNF.25 Indeed, RA is a key physiological factor involved in the induction of α4β7 integrin on lymphocyte surfaces, which is targeted by VDZ.18,26,27 It has also been reported that CD14+ macrophages from the intestinal mucosa of patients with CD are capable of generating RA, which might increase the inflammatory phenotype of these cells.28 During maintenance therapy, a VTL > 19 µg/mL was associated on multivariate analysis with clinical remission, which agreed with previous data from the literature.9,10 However, quartile analyses underlined the lack of specificity of VTL for the diagnosis of clinical remission. Measurement of albumin in blood was not performed in our study, but in contrast to anti-TNF, the impact of serum albumin levels on VTL remains unclear.29,30 The median weight of the included patients was 63 kg [range 50–88], and it was not associated with clinical response on univariate analysis. This could be due to the absence in our cohort of patients with ‘extreme weight’ [>120 kg], since overweight has been identified as a potential clinically important predictor of increased clearance of VDZ.30 Our study has some limitations. The number of included patients was rather low, which precluded meaningful analyses of subgroups, in particular for quartile analyses, evaluation of treatment optimization, and difference between CD and UC. Likewise, the potential diagnostic value of low RA concentrations for clinical remission could not be assessed from our study due to its lack of statistical power. The short duration of maintenance therapy and the retrospective nature of the study could also be pointed out as limits. However, the included patients had a clinical assessment of the disease activity reported prospectively, with objective parameters such as fCal and endoscopic evaluation at the same time decreasing this limitation. Management of patients was based on these clinical scores and independent of the results of VTL, sMAdCAM-1, and RA concentrations. Our work should be considered as a proof-of-concept study, and the analyzed biomarkers can be easily measured in the future with high reproducibility and feasibility in clinical practice. In an ongoing prospective study, we are measuring sMAdCAM-1 and RA concentrations in order to assess their predictive value for clinical remission during VDZ therapy. Awaiting further studies, we can only propose with caution the following strategy. Before starting VDZ therapy, RA concentrations should be measured to help decision-making during the follow-up. Pharmacokinetics combining values of VTL and sMAdCAM-1 should be used if there is a loss of response during maintenance therapy: in the case of a ‘favorable’ pharmacokinetic profile [undetectable sMAdCAM-1 and a VTL of >19 µg/mL], the switch to therapeutic class other than VDZ should be recommended. Optimization of treatment should be proposed if sMAdCAM-1 is detectable with low VTL. With other pharmacokinetic profiles [VTL > 19 µg/mL and detectable sMAdCAM-1; VTL ≤ 19 µg/mL and undetectable sMAdCAM-1], it appears necessary to assess the probability of clinical remission under VDZ: if RA before induction therapy is <1.86 ng/mL, this probability should be low, and a switch of therapeutic class should be proposed. In contrast, if RA is >1.86 ng/mL, optimization of VDZ therapy should be tried. At the moment, this algorithm is to be used with caution, and further studies are needed to improve its level of evidence. In conclusion, undetectable sMAdCAM-1 appears strongly associated with clinical remission during VDZ maintenance therapy in IBD patients. sMAdCAM-1 is actually an inflammatory surrogate marker that probably does not interfere with the effectiveness of VDZ. We observe that it is higher in responder patients before treatment, because the expression of sMAdCAM-1 is also correlated with the expression of MAdCAM-1 on vessels. This indicates that the inflammatory process involved is dependent on lymphocyte migration via this system. This surrogate marker [sMAdCAM-1], however, is unique in predicting the response to VDZ, which is not the case for CRP or fCal. During maintenance therapy, the rate of sMAdCAM-1 falls in responder patients, because it ultimately reflects the decrease in inflammation in these patients. This decreased expression of MAdCAM-1 has already been observed in patients receiving anti-TNF induction therapy.14 Both these induction and maintenance observations suggest that sMAdCAM-1 may bind to α4β7 without inhibiting VDZ uptake. In addition, a pharmacokinetic profile combining undetectable sMAdCAM-1 with high VTL might be potentially be an indicator for TDM. Concentrations of RA before induction are predictive of clinical remission under VDZ therapy. Our findings need to be confirmed in further prospective studies that allow the elaboration of a decisional algorithm. Our data also show that the pharmacokinetics of VDZ are quite complicated and involve different actors/molecules. A mechanistic study of the different molecules involved in the efficacy of VDZ seems to be crucial in order to optimize this treatment in IBD patients. Funding None. Conflict of Interest No conflicts of interest were declared. Author Contributions SP*: conception of the study, data collection, and drafting of the manuscript; NW*: statistical analysis, interpretation of data, and drafting of the manuscript; TDB: data collection; AEB, GB, JF, EDT, BF*: interpretation of data, and critical review of the manuscript; SN: data collection, and critical review of the manuscript; XR*: conception and design of the study, analysis of data, drafting of the manuscript, and study supervision. Abbreviations CD Crohn’s disease CI Confidence interval CRP C-reactive protein DCs Dendritic cells fCal Fecal calprotectin HBI Harvey–Bradshaw index IBD Inflammatory bowel disease NPV Negative predictive value OR Odds ratio PPV Positive predictive value RA Retinoid acid RALDH Retinaldehyde dehydrogenase ROC Receiver operating characteristic Se Sensitivity sMAdCAM-1 Soluble mucosal addressin cell adhesion molecule-1 Sp Specificity TDM Therapeutic drug monitoring TNF Tumor necrosis factor UC Ulcerative colitis VTL Vedolizumab trough level. References 1. Feagan BG , Rutgeerts P , Sands BE , et al. Vedolizumab as induction and maintenance therapy for ulcerative colitis . N Engl J Med 2013 ; 369 : 699 – 710 . Google Scholar CrossRef Search ADS PubMed 2. Sandborn WJ , Feagan BG , Rutgeerts P , et al. Vedolizumab as induction and maintenance therapy for Crohn’s disease . N Engl J Med 2013 ; 369 : 711 – 21 . Google Scholar CrossRef Search ADS PubMed 3. Amiot A , Serrero M , Peyrin-Biroulet L , et al. One-year effectiveness and safety of vedolizumab therapy for inflammatory bowel disease: a prospective multicentre cohort study . Aliment Pharmacol Ther 2017 ; 46 : 310 – 21 . Google Scholar CrossRef Search ADS PubMed 4. Bye WA , Jairath V , Travis SPL . Systematic review: the safety of vedolizumab for the treatment of inflammatory bowel disease . Aliment Pharmacol Ther 2017 ; 46 : 3 – 15 . Google Scholar CrossRef Search ADS PubMed 5. Paul S , Del Tedesco E , Marotte H , et al. Therapeutic drug monitoring of infliximab and mucosal healing in inflammatory bowel disease: a prospective study . Inflamm Bowel Dis 2013 ; 19 : 2568 – 76 . Google Scholar CrossRef Search ADS PubMed 6. Roblin X , Rinaudo M , Del Tedesco E , et al. Development of an algorithm incorporating pharmacokinetics of adalimumab in inflammatory bowel diseases . Am J Gastroenterol 2014 ; 109 : 1250 – 6 . Google Scholar CrossRef Search ADS PubMed 7. Williet N , Paul S , Peyrin-Biroulet L , et al. Pharmacokinetics of infliximab and reduction of treatment for inflammatory bowel diseases . Dig Dis Sci 2016 ; 61 : 990 – 5 . Google Scholar CrossRef Search ADS PubMed 8. Williet N , Boschetti G , Fovet M , et al. Association between low trough levels of vedolizumab during induction therapy for inflammatory bowel diseases and need for additional doses within 6 months . Clin Gastroenterol Hepatol 2017 ; 15 : 1750 – 7 . Google Scholar CrossRef Search ADS PubMed 9. Ungaro RC , Jossen J , Phan BL , et al. Higher vedolizumab trough levels associated with remission in inflammatory bowel disease [IBD] during maintenance therapy . Gastroenterology 2017 ; 152 : S384 – 5 . Google Scholar CrossRef Search ADS 10. Yarur AJ , Bruss A , Jain A , et al. Higher vedolizumab levels are associated with deep remission in patients with Crohn’s disease and ulcerative colitis on maintenance therapy with vedolizumab . J Crohns Colitis 2017 ; 11 : S38 . https://doi.org/10.1093/ecco-jcc/jjx002.057. Accessed February 17, 2017. Google Scholar CrossRef Search ADS 11. Rosario M , French JL , Dirks NL , et al. Exposure–efficacy relationships for vedolizumab induction therapy in patients with ulcerative colitis or Crohn’s disease . J Crohns Colitis 2017 ; 11 : 921 – 9 . Google Scholar CrossRef Search ADS PubMed 12. Boden EK , Shows D , Chiorean MV , et al. Integrin α4β7 expression preceding and saturation during vedolizumab therapy correlate with treatment response in inflammatory bowel disease . Gastroenterology 2017 ; 152 : S39 . Google Scholar CrossRef Search ADS 13. Shigematsu T , Specian RD , Wolf RE , et al. MAdCAM mediates lymphocyte–endothelial cell adhesion in a murine model of chronic colitis . Am J Physiol Gastrointest Liver Physiol 2001 ; 281 : G1309 – 15 . Google Scholar CrossRef Search ADS PubMed 14. Biancheri P , Di Sabatino A , Rovedatti L , et al. Effect of tumor necrosis factor-α blockade on mucosal addressin cell-adhesion molecule-1 in Crohn’s disease . Inflamm Bowel Dis 2013 ; 19 : 259 – 64 . Google Scholar CrossRef Search ADS PubMed 15. Xu Y , Hunt NH , Bao S . The correlation between proinflammatory cytokines, MAdCAM-1 and cellular infiltration in the inflamed colon from TNF-alpha gene knockout mice . Immunol Cell Biol 2007 ; 85 : 633 – 9 . Google Scholar CrossRef Search ADS PubMed 16. Vermeire S , Sandborn WJ , Danese S , et al. Anti-MAdCAM antibody [PF-00547659] for ulcerative colitis [TURANDOT]: a phase 2, randomised, double-blind, placebo-controlled trial . Lancet 2017 ; 390 : 135 – 44 . Google Scholar CrossRef Search ADS PubMed 17. Mwanza-Lisulo M , Kelly P . Potential for use of retinoic acid as an oral vaccine adjuvant . Philos Trans R Soc Lond B Biol Sci 2015 ; 19 : 370 [1671] . 18. Iwata M , Hirakiyama A , Eshima Y , et al. Retinoic acid imprints gut-homing specificity on T cells . Immunity 2004 ; 21 : 527 – 38 . Google Scholar CrossRef Search ADS PubMed 19. Hammerschmidt SI , Friedrichsen M , Boelter J , et al. Retinoic acid induces homing of protective T and B cells to the gut after subcutaneous immunization in mice . J Clin Invest 2011 ; 121 : 3051 – 61 . Google Scholar CrossRef Search ADS PubMed 20. Saurer L , McCullough KC , Summerfield A . In vitro induction of mucosa-type dendritic cells by all-trans retinoic acid . J Immunol 2007 ; 179 : 3504 – 14 . Google Scholar CrossRef Search ADS PubMed 21. Mora JR , Iwata M , Eksteen B , et al. Generation of gut-homing IgA-secreting B cells by intestinal dendritic cells . Science 2006 ; 314 : 1157 – 60 . Google Scholar CrossRef Search ADS PubMed 22. Fransén K , Franzén P , Magnuson A , et al. Polymorphism in the retinoic acid metabolizing enzyme CYP26B1 and the development of Crohn’s disease . PloS One 2013 ; 8 : e72739 . Google Scholar CrossRef Search ADS PubMed 23. Stopfer P , Obermeier F , Dunger N , et al. Blocking lymphotoxin-beta receptor activation diminishes inflammation via reduced mucosal addressin cell adhesion molecule-1 [MAdCAM-1] expression and leucocyte margination in chronic DSS-induced colitis . Clin Exp Immunol 2004 ; 136 : 21 – 9 . Google Scholar CrossRef Search ADS PubMed 24. Wacleche VS , Chomont N , Gosselin A , et al. The colocalization potential of HIV-specific CD8+ and CD4+ T-cells is mediated by integrin β7 but not CCR6 and regulated by retinoic acid . PloS One 2012 ; 7 : e32964 . Google Scholar CrossRef Search ADS PubMed 25. Atreya R , Neumann H , Neufert C , et al. In vivo imaging using fluorescent antibodies to tumor necrosis factor predicts therapeutic response in Crohn’s disease . Nat Med 2014 ; 20 : 313 – 8 . Google Scholar CrossRef Search ADS PubMed 26. Eksteen B , Mora JR , Haughton EL , et al. Gut homing receptors on CD8 T cells are retinoic acid dependent and not maintained by liver dendritic or stellate cells . Gastroenterology 2009 ; 137 : 320 – 9 . Google Scholar CrossRef Search ADS PubMed 27. Mora JR , Bono MR , Manjunath N , et al. Selective imprinting of gut-homing T cells by Peyer’s patch dendritic cells . Nature 2003 ; 424 : 88 – 93 . Google Scholar CrossRef Search ADS PubMed 28. Sanders TJ , McCarthy NE , Giles EM , et al. Increased production of retinoic acid by intestinal macrophages contributes to their inflammatory phenotype in patients with Crohn’s disease . Gastroenterology 2014 ; 146 : 1278 – 88 . Google Scholar CrossRef Search ADS PubMed 29. Shivashankar R , Aberra F , Mendoza Ladd AH , et al. Effect of serum albumin levels on efficacy of vedolizumab in patients with Crohn’s disease . Gastroenterology 2017 ; 152[ Suppl. 1]: s407 – 8 . Google Scholar CrossRef Search ADS 30. Rosario M , Dirks NL , Gastonguay MR , et al. Population pharmacokinetics–pharmacodynamics of vedolizumab in patients with ulcerative colitis and Crohn’s disease . Aliment Pharmacol Ther 2015 ; 42 : 188 – 202 . Google Scholar CrossRef Search ADS PubMed Copyright © 2018 European Crohn’s and Colitis Organisation (ECCO). Published by Oxford University Press. All rights reserved. For permissions, please email: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)

Journal

Journal of Crohn's and ColitisOxford University Press

Published: Sep 1, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off