Review Article

Predictors of Response to Vedolizumab and Ustekinumab in Inflammatory Bowel Disease

A. Barré; J.-F. Colombel; R. Ungaro


Aliment Pharmacol Ther. 2018;47(7):896-905. 

In This Article


Predictors of Response to Vedolizumab in CD and UC

Clinical disease characteristics. Disease activity at baseline seems to be an independent predictor of response to vedolizumab in both UC and CD (Table 1). In subgroup analyses of GEMINI 1 and 2 trials, baseline Mayo score < 9 and CDAI score ≤ 330 were associated with higher clinical remission rates at 6 and 54 weeks when compared to placebo.[1,2] In several observational cohorts, disease activity was also an independent predictor of response. For example, a French multicentre cohort (Groupe d'Etude Thérapeutique des Affections Inflammatoires du tube Digestif [GETAID]), included 294 patients (173 CD and 121 UC), found that patients with a Harvey–Bradshaw index (HBI) score > 10 or a Mayo score > 9 at baseline were less likely to achieve steroid–free clinical remission at weeks 14[6] and 54.[13] In addition, a German cohort including 97 active CD patients, identified a low HBI score and no hospitalisation in the past 12 months as independent predictors of clinical remission at week 14.[5] In the Israeli real–world experience with 130 CD patients, mild clinical activity was associated with an increased likelihood of achieving clinical remission at week 14.[7]

Among others factors studied, subgroup analyses of GEMINI 2 showed higher clinical remission rates in men, younger patients (age < 35 years) and those with a shorter duration of disease ( < 7 years) in CD.[1] In the VICTORY consortium cohort, which included 212 CD patients at 7 sites across the United States, patients with baseline severe disease activity, smoking history and active perianal disease were less likely to achieve clinical remission during maintenance therapy.[23]

The literature on predictors to anti–TNF is consistent with these results. Disease severity in UC and complicated phenotype in CD have been reported as negative predictors of response to anti–TNF agents.[24,25] In contrast, short disease duration in CD before initiation of anti–TNF[26,27] and younger age have been associated with better responses.[28,29]

Inflammatory biomarkers. Contrary to what has been shown with anti–TNF in the SONIC and PRECISE trials,[30,31] higher levels of systemic and bowel inflammation appear to be negative predictors of clinical response or remission to vedolizumab. A prospective study including 172 patients (107 CD and 59 UC) at 2 large academic centres in Boston found that baseline C–reactive protein (CRP) elevation was negatively associated with week 14 response or remission.[4] Amiot et al showed that UC patients with an elevated CRP and leukocytosis were less likely to achieve steroid–free clinical remission at week 14 and during maintenance therapy respectively.[6,13] A national study based on the Swedish National Quality Registry for IBD (SWIBREG), a large prospectively maintained national quality registry from 45 hospitals, was conducted to describe the patient population that was treated with vedolizumab to assess the long–term drug continuation rate and to identify predictors of drug discontinuation.[11] In total, 246 patients were included (147 CD and 92 UC). Elevated CRP at initiation was associated with a higher risk for vedolizumab discontinuation due to lack or loss of response. Finally, a lower CRP or faecal calprotectin at week 14, as compared to baseline predicted clinical remission at week 54 in a large German cohort.[12]

Initial response to vedolizumab. Patients who achieved early response to vedolizumab appear to be more likely to have a long–term response. In the GETAID cohorts, patients who had a clinical response at week 6 were more likely to achieve steroid–free clinical remission at week 14 and 54 in both CD and UC.[6,13] In addition, response and remission status at week 14 predicted clinical remission at week 54 in German cohort.[12]

Early clinical response and mucosal healing with anti–TNF have also been found as positive predictors for long–term response.[32–34] Mucosal healing should be evaluated as a predictor of response in future studies with vedolizumab.

Concomitant therapies. In French cohorts (GETAID), patients with concomitant steroid use at time of vedolizumab induction were less likely to achieve steroid–free clinical remission at week 14 in CD patients[6] and at week 54 for CD and UC.[13] Stallmach et al found that steroid use for less than 25% of the time within the last 6 months predicted a higher likelihood of clinical remission at week 54 in UC.[12]

During the maintenance phase, the Boston cohort showed that the addition of an immunomodulator after induction in CD was a significant predictor of clinical response or remission to vedolizumab at week 54.[10] However, this was not found in other studies. In a population pharmacokinetic and pharmacodynamic analysis, use of concomitant immunosuppression did not affect vedolizumab clearance or concentration, which differs from effects seen with anti–TNF.[35] Indeed, concomitant treatment with immunomodulator with anti–TNF was associated with improved trough levels, reduced production of antibodies infliximab and improved clinical outcomes.[30,36] In addition, in the real–world VICTORY cohort where 54% of patients were treated with concomitant immunomodulator and/or steroids, no differences were noted in terms of clinical response or remission compared to vedolizumab alone.[23] The question of whether to add or continue an immunomodulator in combination with vedolizumab therapy remains uncertain.

Previous anti–TNF exposure. Subgroups analyses of GEMINI 1 and 2 showed higher remission rates in patients with no prior anti–TNF failure in UC and CD.[1,2] Two post hoc analyses from the GEMINI studies reported the efficacy of vedolizumab in UC and CD patients based on prior anti–TNF history.[37,38] Vedolizumab demonstrated a consistent benefit for inducing and maintaining clinical response and remission in both TNF–naïve and TNF–failure patients over placebo. However, rate of response and remission were numerically higher in CD patients receiving vedolizumab as a first biological agent than in patients who had experienced TNF failure (48.4% and 26.6% vs 39.7% and 21.8% at week 10). These higher rates persist at week 52.[38] The same results were found in UC patients, where patients naive to anti–TNF had higher rates of response (absolute difference [AD]: 15.5%) than patients with anti–TNF failure (AD: 7.0%) compared to placebo at week 6, but the AD in remission rates observed was the same in the maintenance therapy.[37]

In the real–world VICTORY cohort, previous anti–TNF exposure was associated with lower rates of clinical remission and mucosal healing in CD patients.[23] Similarly, in UC patients, the VICTORY consortium observed that prior anti–TNF exposure was associated with a decreased chance of clinical response and remission during maintenance therapy.[39] Stallmach et al showed that patients who were naïve to anti–TNF were more likely to achieve clinical remission at week 54 only in UC patients.[12] Recently, a clinical prediction model and tool for vedolizumab therapy in CD was developed with GEMINI 2 participants and then validated in the VICTORY cohort. Predictors included in this tool were the absence of prior exposure to anti–TNF (+4 points), the absence of prior CD–related hospitalisation in the past 12 months (+6 points), the absence of prior fistulising disease (+3 points) and baseline CRP concentration (−0.1 points per mg/L). Patients with score > 8 points were twice as likely to achieve clinical remission, corticosteroid–free remission and mucosal healing after 26 weeks of vedolizumab therapy. This model suggests that the ideal positioning of vedolizumab in CD patients should be prior to anti–TNF exposure (used as a first–line biological agent) and in an uncomplicated disease.[40]

However, these results are similar to those seen in anti–TNF where response rates are consistently better in patients naive of biological agents. Results of comparative effectiveness studies between vedolizumab and anti–TNF will be essential to determine the precise positioning of vedolizumab in the treatment algorithm for IBD.[41]

Pharmacological aspects. There is a well–established correlation between serum trough levels of anti–TNF and clinical response.[42,43] In the same way, post hoc analyses of GEMINI study data revealed a positive relationship between vedolizumab exposure and efficacy. Higher vedolizumab concentrations were associated with higher clinical remission rates at week 6.[44] The exposure–efficacy relationship was steeper for UC than CD, when comparing results from GEMINI 1 and 2. This relationship was affected by several factors, including previous anti–TNF use, baseline albumin concentration and CRP. The remission probability was approximately 10% higher in TNF–naïve patients. In addition, in a prospective study from 2 French hospitals that included 47 patients with CD or UC receiving induction therapy with vedolizumab, low trough levels of vedolizumab at week 6 ( < 19.0 mg/mL) were associated with need for additional doses (given at week 10 and then every 4 weeks).[45] All patients receiving these additional doses achieved a clinical response 4 weeks later. Other prospective studies are needed to assess vedolizumab dose individualisation and optimisation particularly during maintenance.

Two factors have been identified to influence anti–TNF serum concentration, and are negatively correlated with response to anti–TNF: low serum albumin and obesity.[46,47] A population pharmacokinetic and pharmacodynamic analysis of vedolizumab showed that very low albumin concentrations and high body weight values were potentially clinically important predictors of clearance of vedolizumab.[35] However, these factors have not been evaluated in observational studies with vedolizumab.

Finally, immunogenicity rates, with development of antibodies against vedolizumab, appear to be very low ( < 5%),[35,45] so response is less likely to be related to immunogenicity, contrary to what is described with anti–TNF.[48–50]

Gut microbiome. A recent study using a prospectively recruited cohort of patients with IBD initiating vedolizumab therapy was performed to define the relationship between gut microbial taxonomic composition and function and response to vedolizumab therapy.[51] Eighty–five patients (42 CD and 43 UC) were included. Community alpha diversity at baseline was significantly higher in CD patients who achieved clinical remission at week 14. Two taxa species, Roseburia inulinivorans and Burkholderiales, were significantly more abundant at baseline in week 14 remitters. Conversely, beta diversity was lower at baseline among those achieving remission. In addition, changes in microbial function may be associated with response. Specifically, 13 pathways including branched chain amino acid synthesis were significantly enriched in CD patients achieving week 14 remission. No significant association was found between gut microbiota and clinical remission at week 14 in UC patients. This study concluded that gut microbiome at baseline could be an indicator of clinical response. However, further validation and feasibility studies are needed prior to translating these findings into clinical practise.

Predictors of Response to Ustekinumab in CD

Clinical disease characteristics. Subgroups analyses in the UNITI trials showed higher clinical response rates in female patients, younger patients, patients with a shorter disease duration ( ≤ 5 years) and those with ileocolonic disease at week 6 and 44 in ustekinumab group compared to placebo.[15] Various observational studies have been conducted looking at predictors of response to ustekinumab (Table 2). A retrospective observational cohort study performed at 2 academic tertiary care institutions in Canada included 167 CD patients treated with ustekinumab assessed clinical factors predictive of response at 6 months. In multivariable analysis, patients with ileocolonic disease were more likely to have clinical response at 6 months. In contrast, patients with an HBI score ≥ 7 at induction were less likely to achieve clinical response, as were patients with stricturing disease.[16] The same group performed a retrospective cohort study of 104 CD patients who responded to ustekinumab induction and continued long–term maintenance.[20] Colonic and ileocolonic disease were associated with lower risk for loss of response during maintenance therapy. In contrast, HBI score ≥ 7 at induction and stricturing phenotype were associated with a higher rate of loss of response. In a retrospective open–label study across 42 Spanish tertiary IBD centres analysing 116 CD patients who were resistant to at least 1 conventional immunosuppressant and 1 TNF antagonist, no factors predicted the initial clinical response to ustekinumab during induction.[19] However, a history of previous intestinal resection was associated with an increased risk of loss of response to ustekinumab during the maintenance phase.

Similar predictors have also been observed in the literature on anti–TNF therapy. Isolated colonic disease was associated with an improved response to anti–TNF in CD, in contrast to previous surgery and stricturing or penetrating phenotype.[25,52]

Initial response to ustekinumab. In a Spanish cohort, the initial response to ustekinumab was associated with clinical benefit at the end of follow–up at a median of 10 months (interquartile range: 5–21).[19] This finding is consistent with data from the CERTIFI study.[14]

Pharmacological aspects. A prospective study in 2 academic centres in Canada investigated associations between trough concentrations of ustekinumab and clinical, biomarker and endoscopic outcomes.[53] Sixty–two CD patients, who were anti–TNF refractory or intolerant, were enrolled. Patients received 90 mg of ustekinumab subcutaneously at weeks 0, 1 and 2 during induction and 90 mg every 4 or 8 weeks during maintenance. The mean trough concentration of ustekinumab at 26 weeks was higher in patients with an endoscopic response (defined by SES–CD reduction ≥ 50%). An optimal ustekinumab threshold trough concentration at week 26 or later was found to be 4.5 μg/mL (area under the curve = 0.67). Maintenance trough concentrations of ustekinumab above 4.5 μg/mL at 26 weeks or later were associated with endoscopic response. No association was found with clinical outcomes. It is important to note that the induction regimen was different than the currently approved dosing for CD.

Similar to vedolizumab, immunogenicity rates with ustekinumab use appear to be low: 2.3% at 1 year in the UNITI studies[15] and 0% at 26 weeks in the Canadian experience.[53]

Concomitant immunomodulator. In a retrospective observational study performed in tertiary French (GETAID) and Swiss centres, 122 CD patients refractory to anti–TNF therapy were studied. Concomitant use of immunosuppressant at baseline was a predictive factor of clinical benefit at 3 months.[18] Ma et al found that concurrent immunomodulator was associated with a reduced risk of loss of response during maintenance treatment but not for induction.[16,20] However, Battat et al showed similar rates of steroid–free clinical remission, endoscopic response and remission between patients with or without immunomodulator. In addition, the average ustekinumab concentrations were similar regardless of immunomodulator use at weeks 10 and 26.[53] Further studies are needed to understand the potential benefit of concomitant immunomodulator with ustekinumab.

Previous anti–TNF or immunomodulator exposure. In phase 3 clinical trials, the rates of clinical response and remission at induction were higher in patients who had failed or were intolerant to conventional therapy (majority naïve to anti–TNF; UNITI–2) than in patients who had failed anti–TNF therapy (UNITI–1).[15] There is no data in observational studies comparing the response to ustekinumab between anti–TNF naive and exposed patients, since all patients included in these studies had failed or were intolerant to anti–TNF therapy.

Regarding prior immunomodulator exposure, a Spanish cohort found that having previously used 2 or more immunosuppressive drugs was associated with a long–term clinical benefit during maintenance (median follow–up: 10 months).[19]