Targeting S1P Receptors, A New Mechanism of Action for Inflammatory Bowel Disease Therapy

Current therapeutic approaches to inflammatory bowel disease have evolved toward target-specific treatments that are being discovered through the continuously expanding recognition of the immunologic pathways that dominate disease pathogenesis. Among such pathways, focus has been placed on the anti-inflammatory effect of preventing inflammatory cell traffic to the affected intestine, most often of lymphocytes, the main cellular effectors in all chronic inflammatory processes. The first successful clinical application of this translational approach occurred in 1998 with the approval of natalizumab, followed in 2014 by that of vedolizumab, a monoclonal antibody against the α4β7 integrin, in both UC and Crohn’s disease (CD).

Through their study, Sandborn et al sought to further expand the field of antitraffic therapies by studying the effects of a first-in-kind orally administered sphingosine-1-phosphate (S1P) receptor 1 and 5 (S1P1, 5-selective) agonist (RPC1063/ozanimod) in patients with ulcerative colitis (UC). S1P is a pleiotropic sphingolipid with diverse biological and immunologic functions, which signals through 5 G-protein–coupled receptors (S1P1–5; Annu Rev Biochem 2009;78:743–768). S1P mediates the traffic of lymphocytes from lymphoid organs into the lymph, following a low to high concentration gradient. Ozanimod is a small molecule agonist that binds to S1P1/5, inducing internalization and eventual degradation of the receptors (functional antagonism), thus blocking the egress of lymphocytes, which remain trapped within lymph nodes.

Sandborn et al conducted a randomized, double-blind, phase II clinical trial in patients with UC, who received one of 2 doses of ozanimod (0.5 or 1 mg) or placebo orally, as induction and medium-term (32 weeks) maintenance therapy. Patients were recruited at 57 centers in 13 countries. All patients had moderate to severe UC as defined by a Mayo Clinic score of 6 to 12, with all patients having active endoscopic inflammation confirmed via blinded centrally read endoscopy (Mayo endoscopic subscore of 2 or 3). No differences existed between the 3 groups regarding concomitant medications (glucocorticoids, 37%-40%; aminosalicylates, 79%-88%) or previous medication use (immunosuppressants, 26%-37%; tumor necrosis factor antagonists, 15%-20%).

The primary outcome was clinical remission (Mayo Clinic Score ≤ 2; no subscore > 1) at week 8. This was achieved by 11 of 67 patients (16%) and in 9 of 65 patients (14%) who received 1 or 0.5 mg of ozanimod, respectively. In contrast, only 4 of 65 patients (6%) who received placebo achieved remission, the difference being statistically significant for the 1-mg (P = .048) but not the 0.5-mg group (P = 0.14). Patients who received 1 mg of ozanimod had significantly better secondary outcomes. These included clinical response at week 8 (placebo, 24/65 [37%]; 0.5 mg ozanimod, 35/65 [54%, P = .06]; 1 mg ozanimod, 38/67 [57%; P = .02]), mucosal healing at week 8 (placebo, 8/65 [12%]; 0.5 mg ozanimod, 18/65 [28%, P = .03]; 1 mg ozanimod, 23/67 [34%; P = .002]); and, histologic remission, defined as a Geboes score of <2 (placebo, 7/65 [11%]; 0.5 mg ozanimod, 9/65 [14%, P = .63]; 1 mg ozanimod, 15/67 [22%; P = .07]). Among patients who achieved clinical remission at week 8 and were followed to 32 weeks, 2 of 4 patients in the placebo group, 7 of 9 in the 0.5 mg ozanimod group, and 5 of 11 in the group of 1 mg ozanimod were still in remission.

As expected by its mechanism of action, ozanimod administration resulted in significant decrease in the number of circulating lymphocytes in both treatment groups. Severe decreases in lymphocyte counts (grade 3) were reported in 9 of 67 patients in the group that received 1 mg of ozanimod. There were no differences noted between the 3 groups in the most common adverse events. Adverse events of specific interest for S1P modulation included 1 case of asymptomatic first-degree atrioventricular block with worsening of preexisting bradycardia, 4 cases with a >3-fold increase in serum alanine aminotransferase, and 1 case of squamous cell carcinoma of the skin in a patient who had also received mercaptopurine for >2 years.

Comment.

Current advances in our understanding of the pathogenesis of UC and CD have clearly demonstrated that these are complex diseases with endogenous clinicopathologic variability. This reflects the great immunogenetic diversity of study subjects, with diverse genetic defects and overlapping immunologic pathways, which in concert with environmental factors (eg, diet and microflora) culminate into the phenotype of an autonomous dysregulated immune response. The translational effect of such variability is that no target-specific therapy has achieved a universal response in every patient. In fact, clinical remission and response rates have fallen far below the level of 50% in all major trials.

The ozanimod trial was no exception. Although the administration of high-dose ozanimod demonstrated a significant benefit over placebo, the percentage of patients (16%) who achieved the primary endpoint of clinical remission at week 8 could be best categorized as underwhelming. Its modest effect could have been easily missed, if not for the excellent study design, which enrolled patients with confirmed inflammation and relied on objective centralized readouts. Nevertheless, in addition to the aforementioned immunogenetic heterogeneity of the participants, certain study-related factors may account for the modest effect. Most important, 8 weeks may have been too early of a time point for ozanimod’s mechanism of action. A major lesson learned from prior traffic-targeted therapies, natalizumab and vedolizumab experiences, is that these therapies may demonstrate slower responses, with maximum benefit observed as late as by week 14 or even later, according to recent post hoc analyses of the GEMINI trials. From a pathogenesis perspective, a longer time may be required to deplete tissue-resident T-cell populations within the intestine. This is further supported clinically by the fact that more patients who received either dose ozanimod were in remission by week 32, than by week 8. Furthermore, problems inherent to multicenter trials, such as intercenter and regional variability may influence the overall selection and response characteristics. Finally, the relatively small number of patients included in each arm and the considerable percentage of difficult to treat, immunosuppressant with or without anti–tumor necrosis factor-resistant cases may have led to underestimation of the true effect of the drug. In any case, the significantly higher remission rates induced by ozanimod further validates the efficacy of therapies that interfere with inflammatory cell traffic in UC. Although lymphocytes (both B and T cells) seem to be the primary target of S1P agonists, we have identified diverse cell populations in mouse models, such as dendritic (DC) and endothelial cells, as additional cellular targets (Mucosal Immunol 2016 Apr 6. doi: 10.1038/mi.2016.35). Thus, apart from a lymphopenic effect, S1P receptor agonists may also affect DC migration and vascular barrier function, further diversifying their anti-inflammatory properties and targeting not only adaptive, but also innate immunity.

Safety issues are always a major concern during drug development and clinical application. Adverse effects of S1P receptor modulators are dictated by the binding properties of each drug and the tissue distribution of the receptors blocked (Nat Rev Drug Discov 2009;8:297–307). From that perspective, ozanimod has a safer predicted profile than the prototype promiscuous S1P receptor agonist fingolimod (FTY720/Gilenya), which is approved for multiple sclerosis. Whereby fingolimod binds to all but 1 (S1P2) S1P receptor (S1P1-5), ozanimod interacts only with 2 (S1P1 and S1P5). Ozanimod has also a much shorter half-life, which leads to a much faster rebound of peripheral lymphocyte counts upon removal of the drug from circulation. Atrial myocytes express S1P receptors, thus a feared complication for both drugs is cardiac dysrhythmias, in particular bradycardia and atrioventricular block, which calls for close supervision of patients, particularly those with a history of related problems. In contrast, S1P5 is expressed in the central nervous system and S1P1 on cells that migrate to central nervous system; thus, cases of progressive multifocal leukoencephalitis have been reported in 3 patients to date treated with fingolimod. Notably. however, some had also received natalizumab. Finally, although ozanimod-induced lymphopenia is likely to be primarily responsible for the anti-inflammatory effect of the drug, by itself is also a potential reason for concern. In the study by Sandborn, 13% of the patients developed stage 3 lymphopenia, yet interestingly serious infections were not reported. As the follow-up of patients was short, no conclusions can be drawn on the long-term safety of ozanimod, until larger studies with longer observation are completed.

There are certain advantages for the use of oral medications over parenteral drugs. Oral administration is convenient, because it involves no discomfort from injection or clinic visits for intravenous administration. Furthermore, orally administered small drugs are not immunogenic, which stands in sharp contrast with monoclonal antibodies that elicit antidrug antibodies and hampers their efficacy. Another advantage is their short half-life (usually hours or days; 19 hours for ozanimod), which allows for rapid elimination (ie, in the event of infection), something that is not achievable with monoclonal antibodies that have weeklong half-lives. Finally, at least theoretically, small drugs have lower manufacturing costs owing to the relatively simpler production process. By contrast, patients on oral medications may show lower adherence to their treatment. In addition, it is considered that target specificity might be better achieved with monoclonal antibodies than with small drugs.

The final verdict for ozanimod will await completion of phase III trials. Whether this drug will find a position in future treatment algorithms for UC will in our opinion depend on 2 primary factors. First, its efficacy should align with the current therapeutic paradigm, which calls not only for symptom resolution but also for endoscopic healing and the histologic absence of inflammation. The drug seems to be on track to accomplish this, at least for UC. Second, no new safety red flags should be raised during upcoming trials. In this context, a recent survey among patients with inflammatory bowel disease reported their willingness to accept considerable risks for medication-induced complications with the aim to maintain disease control (Am J Gastroenterol 2015;110:1675–1681). Its future in CD is more uncertain. Trial design limitations based on subjective Crohn’s Disease Activity Index scoring add to the uncertainty of whether a drug is indeed useless or whether the trial design may be to blame for past and recent failures (eg, tofacitinib and anti–MAdCAM-1 antibodies in CD). We learn little from poorly designed CD trials and do not need ozanimod or any other drug to join what Sandborn refers to as “the graveyard,” owing to a flawed randomized controlled trial design. Indeed, Novartis got into the S1P field for UC years earlier with KRP203, a drug that seems to be remarkably similar to ozanimod. Their trial, with a very small sample size, was unable to produce a signal and was abandoned early. Finally, it is inevitable that, with a constantly expanding therapeutic armamentarium for UC and CD, a shift toward personalized medicine will become of paramount importance. In such scenario, clinicopathologic characteristics will be linked with genetic, immunologic, and environmental fingerprints in individual cases, which will largely dictate the selection of the most appropriate target-specific therapies of the future.