A Clinical Decision Support Tool May Help to Optimise Vedolizumab Therapy in Crohn's Disease

Parambir S. Dulai; Aurelien Amiot; Laurent Peyrin-Biroulet; Vipul Jairath; Melanie Serrero; Jerome Filippi; Siddharth Singh; Benjamin Pariente; Edward V. Loftus Jr; Xavier Roblin; Sunanda Kane; Anthony Buisson; Corey A. Siegel; Yoram Bouhnik; William J. Sandborn; Karen Lasch; Maria Rosario; Brian G. Feagan; Daniela Bojic; Caroline Trang-Poisson; Bo Shen; Romain Altwegg; Bruce E. Sands; Jean-Frederic Colombel; Franck Carbonnel


Aliment Pharmacol Ther. 2020;51(5):553-564. 

In This Article


Although the introduction of biologic therapy has greatly improved the management of CD, fewer than one-third of patients treated with a TNF antagonist, VDZ, or ustekinumab achieve corticosteroid-free clinical remission 1 year following initiation of therapy. The reasons for this unsatisfactory circumstance are complex and include disease heterogeneity, delayed initiation of therapy, drug sensitisation, development of disease-related complications and adverse effects of medical therapy. However, irrespective of specific causes, low efficacy rates translate into poor incremental cost-effectiveness estimates and reluctance by payers to fund these therapies. One of the fundamental concepts of personalised medicine is identification of patients who are more likely to respond to a specific therapy. Accordingly, we previously developed and validated a CDST for predicting response to VDZ therapy in CD. In the current study, we extend the clinical utility of this CDST through several additional analyses and an international collaboration between the VICTORY consortium and GETAID investigators, which also benefited from access to the phase 3 GEMINI trial data.

In the GEMINI 2 trial dataset, the CDST-defined response categories predicted highly significant differences in measured VDZ exposure throughout the 52-week study, and this was observed irrespective of whether VDZ exposure was measured at trough (pre-dose), peak (2-hour post-dose) or midway between infusions. Most notably, the measured week 6 trough VDZ concentrations in the CDST-defined low- and high-probability response groups from the GEMINI 2 cohort had non-overlapping interquartile ranges, with a twofold higher median concentration being observed in the high-probability group. In the GEMINI 2 trial dataset, we also observed that the CDST-defined response categories predicted highly significant differences in rapidity of onset of action and absolute reductions in HBI throughout the 52-week study. Specifically, throughout the first 14 weeks of VDZ treatment, the high-probability response group had at least a twofold increased reduction in HBI from baseline compared with the low-probability response group (P < 0.001). Together these observations demonstrate a relationship between CDST-defined probability groups, VDZ exposure, and rapidity of onset of action and an ability to identify patients undergoing VDZ therapy who will have low drug exposure and slower onset of action, and therefore may benefit from early dose intensification.

In the VICTORY and GETAID cohorts, we observed the CDST to predict differences in VDZ response comprising both rapidity of onset of action and overall effectiveness as defined by both symptomatic and endoscopic remission rates. In the VICTORY cohort, rates of endoscopic remission were significantly different between probability groups as defined by the CDST, and this predicted difference between CDST-defined response groups remained significant even after accounting for factors known to influence disease outcomes. In the GETAID cohort, rates of clinical remission at week 14 were significantly different between the CDST-defined low- and intermediate- or high-probability groups, and we observed significant differences in rapidity of onset of action between these groups as measured by reductions in HBI through 14 weeks of therapy. Using the GETAID cohort, we expanded on our observations from the GEMINI 2 cohort by observing that these differences in onset of action between the CDST-defined low and intermediate or high probability of response groups appeared to be overcome by administration of a week 10 infusion. Using the VICTORY cohort, we also observed that only the low and intermediate probability of response groups benefited from VDZ interval shortening for nonresponse. Collectively, these findings provide further evidence of the validity of the CDST across diverse CD patient populations and its potential application to identify CD patients starting VDZ therapy who could potentially benefit from early dose optimisation through a week 10 infusion and/or shortening of VDZ maintenance intervals. However, this strategy requires further validation in prospective studies. It is also important to clarify that CD labelling in the European Union includes administration of an additional week 10 dose of VDZ; however, this is not authorised in labelling in North America, including the United States.

Another important finding of the study was that the CDST seemed to effectively predict "hard endpoints" such as endoscopic remission and CD-related surgery. Although other tools have been developed to prognosticate overall risk of complications among CD patients to help guide patient discussions for starting biologics,[18,19] it is difficult to know who remains at risk for disease-related complications after biologic initiation.[20] We observed the low and intermediate CDST groups of VDZ therapy patients to be significantly less likely to achieve endoscopic remission and more likely to undergo surgery relative to the high CDST group. Given the observations made for variability in measured VDZ exposure, onset of action, response to interval shortening for the low-probability patients, and endoscopic remission across CDST groups, it could be hypothesised that VDZ dose optimisation in low and intermediate CDST groups could be effective in off-setting the increased risk for surgery observed by optimising the achievement of endoscopic remission, an endpoint associated with risk of disease-related complications. This concept, however, requires formal assessment in a well-designed phase 3 trial similar to REACT,[21] and it is worth noting that despite a significant difference in endoscopic remission rates between the low and intermediate CDST groups, no significant difference was observed for risk of surgery between these groups.

Our study has several strengths, including the multinational validation in mixed practice settings and extension of CDST predictions to VDZ exposure, onset of action, response to interval shortening and risk of surgery. Several limitations, however, require acknowledgment. Post hoc analyses of clinical trial datasets have inherent limitations that prevent definitive conclusions, and real-world data have inherent limitations in collection methods and consistency of assessments that may have biased the results. No specific or consistent timing was applied for the assessment of response to interval shortening in the VICTORY consortium, and the physician global assessment was used, which carries a risk for misclassification. The GETAID cohort was an early, treatment-refractory population, and only a subset had all necessary data to calculate the CDST. Therefore, the analyses were limited to a subset of patients, which could still introduce a selection bias, and significance in comparisons was not observed beyond week 14. Further analyses are likely therefore still needed to fully capture the validity of this CDST to assist in treatment optimisation, particularly with regard to the use of a week 10 dose for optimisation. Accordingly, well-designed phase 3 trials focusing on optimisation of disease outcomes and treatment response for VDZ using the CDST as an enrichment or stratification tool could help overcome the current gap in personalised medicine for inflammatory bowel disease.

In conclusion, the previously built CDST for VDZ in CD appears to be valid across multiple cohorts and has significant prognostic and predictive capacity to guide therapeutic decisions in routine practice. Patients deemed low probability for response to VDZ may potentially benefit from a week 10 dose to optimise drug exposure and rapidity of onset of action. When implementing aggressive treat-to-target monitoring strategies, low- or intermediate-probability patients may benefit most from this strategy, and healthcare systems may consider stratified follow-up intervals based on probability of response. Finally, among high-probability patients, we observed rapid onset of action and high drug concentrations. If these patients fail to respond to VDZ, it may be related to an immunologic or genetic mechanism, and further studies are needed to help identify additive biomarkers to further optimise the predictive capacity of the CDST for VDZ.