Treatment options for patients with moderate-severe ulcerative colitis (UC) have increased in the last decade with the availability of six different classes of medications.¹ These medications have diverse mechanisms of action, targeting different pathways of inflammation, including neutralizing cytokines that promote inflammation (tumor necrosis factor-α [TNF] antagonists) or drive the differentiation and function of specialized immune subsets (interleukin [IL]-12/23 antagonists or selective IL-23 antagonists); blocking signal transduction cascades downstream of these pathways (Janus Kinase inhibitors [JAK inhibitors]); or modulating lymphocyte trafficking (anti-α4β7 integrin antibodies such as vedolizumab and sphinosine-1-phosphate [S1P] receptor modulators such as ozanimod and etrasimod).^{2, 3} There is wide inter-individual variability in response to specific medications, and drivers of this heterogeneity of treatment effect are critical to understand to be able to choose the best therapy for each individual patient. For newer advanced therapies, one key biologic driver of this heterogeneity is prior exposure to TNF antagonists.⁴ In clinical trials, certain classes of medications such as lymphocyte trafficking inhibitors appear to have a smaller placebo-adjusted treatment effect in patients with prior exposure to TNF antagonists compared with TNF antagonist-naïve patients, whereas these effects are not readily apparent for other therapies.⁵ It is conceivable that exposure to TNF antagonists may “reprogram” the immune system, altering responsiveness to subsequent therapies in a manner that is dependent on the second drug’s mechanism-of-action.

While absolute rates of achieving clinical remission are higher in TNF antagonist-naïve patients compared with TNF antagonist-exposed treated with advanced therapies, so is the placebo response rate. In order to examine how prior exposure to TNF antagonists modifies placebo-adjusted treatment effect of different classes of advanced therapies, we conducted a systematic review and meta-analysis of phase 2 and 3 RCTs of approved advanced therapies vs. placebo, comparing treatment effect in patients with TNF antagonist-naïve patients vs. those with prior exposure to TNF antagonists, accounting for corresponding placebo remission rates. Since prior exposure to TNF antagonists was a stratification variable in all these trials, they preserved the benefits of randomization in subgroup analysis.

This systematic review was performed using an a priori established protocol and is reported according to Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA).⁶

Our search strategy yielded 5,350 unique studies. From this, we included 17 RCTs of advanced therapies vs. placebo that reported induction of clinical remission, stratified by prior biologic exposure. These included five RCTs of lymphocyte trafficking inhibitors (two RCTs of vedolizumab; three RCTs of S1P receptor modulators),^11–14 six RCTs of anti-interleukins (one RCT of IL12/23 antagonist; five RCTs for selective IL-23 antagonists)^15–20 and six RCTs of JAK inhibitors (two RCTs of tofacitinib; four RCTs of JAK1 inhibitors).^21–24 Details of these trials are shown in Table 1. Overall, these trials included 8871 patients of whom 66% were treated with advanced therapies, and 48% had prior exposure to biologics, >90% of whom were exposed to TNF antagonists with or without exposure to other advanced therapies. U-ACHIEVE, U-ACCOMPLISH, and INSPIRE trials stratified their outcomes by prior advanced therapy without specifying which prior advanced therapies patients received. All trials were deemed to be at low risk of bias.

Biologic drivers of heterogeneity of treatment effect are important to understand for personalized medicine in patients with IBD.²⁵ One consistent adverse prognostic factor for treatment efficacy in patients with UC is prior failure of TNF antagonists. However, prior exposure to TNF antagonists may variably modify treatment efficacy of subsequent advanced therapies. In this systematic review and meta-analysis, we made several key observations by methodically examining the efficacy of different classes of advanced therapies based on prior TNF antagonist exposure status. We confirmed prior observations that exposure to TNF antagonists significantly reduces the efficacy of lymphocyte trafficking inhibitors in inducing remission, including both vedolizumab and S1P receptor modulators, by approximately 50%. In contrast, prior exposure to TNF antagonists was associated with a significant increase in the efficacy of JAK inhibitors in inducing remission, with two-fold higher efficacy in TNF antagonist-exposed vs. TNF antagonist-naïve patients. Prior exposure to TNF antagonist did not modify the efficacy of selective IL-23 antagonists; in contrast, in the UNIFI trial the efficacy of IL-12/23 antagonist, ustekinumab was significantly increased with prior exposure to TNF antagonists. These observations were similar for induction of endoscopic improvement endpoint. Clinically, these findings reassure positioning of JAK inhibitors after exposure to TNF antagonists; while these recommendations from the FDA were driven by safety concerns with JAK inhibitors,²⁶ our findings suggest that prior exposure to TNF antagonists may potentiate the efficacy of JAK inhibitors.

With availability of several different treatment options, optimal therapeutic sequencing is critical to understand. There are limited studies in patients with IBD that have examined treatment sequencing. Observational comparative effectiveness studies in patients with UC have shown that tofacitinib is more efficacious than vedolizumab in patients with prior failure of TNF antagonists.^27–29 However, these same studies have not examined comparative effectiveness of these therapies in TNF antagonist-naïve patients, and importantly, have not been able to discern changes in the effectiveness of vedolizumab or tofacitinib following exposure to TNF antagonists. While trials of therapeutic sequencing are limited in IBD, such trials have been conducted in other immune-mediated diseases, such as rheumatoid arthritis. In the SELECT-COMPARE trial in patients with rheumatoid arthritis with inadequate response to methotrexate, clinical remission was achieved in 29% upadacitinib- vs. 16% adalimumab-treated patients.^{30, 31} At week 26, patients with inadequate response to upadacitinib or adalimumab were switched to the other therapy. After this switch, a significantly higher proportion of patients who switched to upadacitinib after failure of adalimumab achieved clinical remission compared with those who switched to adalimumab after failure of upadacitinib (35% vs. 21%). These published findings, together with the observation in this study that exposure to TNF antagonists augments the efficacy of JAK inhibitors, support the potentially paradigm-shifting concept that exposure to one drug may differentially influence subsequent efficacy of a second drug.

It is critical to understand mechanistic insights into these observations which need further biologic characterization. We propose that exposure to TNF antagonists could durably change the immune system in patients with UC. Indeed, it has been previously demonstrated that infectious exposures can result in lasting consequences to the immune system. For example, a single acute gastrointestinal infection was shown to disrupt the communication between the immune system and the tissue, causing an “immunological scar” that compromised tissue immunity and homeostasis in the long-term.³² Further, non-infectious exposures can have lasting effects on the immune system. Smoking was shown to exert a transient effect on innate immune responses, but an enduring effect on adaptive immunity that persisted well beyond smoking cessation.³³ Thus, the current findings raise the intriguing possibility that exposure to TNF antagonists could result in lasting effects on the immune system that differentially alter responsiveness to therapies with distinct mechanisms of action. TNFα-blockade reduces leukocyte trafficking and decreased leukocyte trafficking may contribute to a reduction in response rates to lymphocyte trafficking therapies.^{34, 35} The possible potentiation of JAK inhibitors in those with prior exposure to TNF antagonists could be due to the IL-6/JAK/STAT pathway that is initially suppressed (~2 weeks) but later increases in some patients starting at week 8.³⁶ Moreover, drugs can cause changes in gene expression that persist long after the drug has been discontinued, through both direct and indirect mechanisms.^{37, 38} In a recent study, using a network transcriptomic approach to predict efficacy of drug targets in in silico trials agrees with the findings we report. In this study, JAK inhibitors were significantly more likely to upregulate a 34-gene signature of molecular remission in patients with prior exposure to TNF antagonists, whereas vedolizumab and S1P receptor modulators upregulated only a small subset of these genes.³⁹ These findings suggest epigenetic rewiring and a foundational impact on gene regulatory networks that impact gene expression after exposure to TNF antagonists.

Another provocative observation in the current study is that exposure to TNF antagonists enhanced subsequent efficacy of IL-12/23 antagonists, but not selective IL-23 antagonists. IL-12 was once considered to be the primary driver of autoimmune and immune-mediated diseases, including IBD, but the discovery of IL-23 led to a paradigm shift whereby IL-23 and Type 17 immunity are now thought to be far more critical.⁴⁰ Nonetheless, IL-12 may still play an important role in the pathogenesis of chronic intestinal inflammation by playing an early role in disease development relative to IL-23, which may play a temporally distinct role in controlling disease chronicity.⁴¹ We speculate that treatment with TNF antagonists may reshape the Type 17 immune-driven nature of UC into a more Type 1 immune-biased disease, providing a potential explanation for enhanced efficacy of therapies that target IL-12. Indeed, TNF is known to inhibit IL-12 production by macrophages;^{42, 43} hence, TNF blockade may lead to de-repressed (increased) IL-12 production, resulting in a Type 1 immunity-dominant microenvironment that is more responsive to IL-12 antagonism.

While the findings are intriguing, there are important limitations that should be considered. First, we combined classes of therapies based on similarity of putative mechanism of action, such as lymphocyte trafficking inhibitors and S1P receptor modulators, even though their exact mechanism of blocking trafficking is different with potential biologic effects. On subgroup analysis by class of advanced therapies, our findings were stable. Second, while we intended to examine the impact of prior exposure to TNF antagonists on heterogeneity of treatment efficacy, for some trials, these results were only available based on prior exposure to any biologic. In these trials, TNF antagonist was the primary biologic exposure in >90% of instances across trials, and in some trials, patients could also have been exposed to additional biologics. Third, among TNF antagonist-exposed patients, there is a subset of patients that have failed two or more types of advanced therapies, which may be negatively associated with therapeutic success.^44,45 Unfortunately, we are unable to examine the impact of exposure to multiple cases of medications on treatment efficacy modification due to limited reporting of these date. Since they are definitively categorized as TNF-exposed, their impact on the analysis results is expected to be minimal. Fourth, we were unable to explicitly adjust for other potential confounding variables. However, since prior exposure to TNF antagonists was a stratification variable in trials, it preserved the benefits of randomization such that patients were randomly assigned to drug or placebo in each stratum.

In conclusion, through a meta-analysis of RCTs, we observed that there is significant heterogeneity of treatment efficacy for induction of remission with different advanced therapies in patients with moderate-severe UC based on prior exposure to TNF antagonists. Prior exposure to TNF antagonists appears to potentiate the efficacy of JAK inhibitors and IL-12/23 antagonists but may attenuate responsiveness to lymphocyte trafficking inhibitors. Future studies on the mechanistic insight for these intriguing observations are warranted.