Lesion remyelinating activity of GSK239512 versus placebo in patients with relapsing-remitting multiple sclerosis: a randomised, single-blind, phase II study

Multiple sclerosis (MS) is a chronic progressive neurologic disorder, characterised by inflammatory-demyelinating lesions in the central nervous system (CNS) and associated axonal loss. Relapsing-remitting MS (RRMS) accounts for 85% of all incident cases of MS and is associated with episodes of CNS dysfunction, reflecting acute inflammatory focal demyelination and subsequent spontaneous recovery, in turn reflecting resolution of focal inflammation and remyelination, with a variable contribution of CNS plasticity [1]. Remyelination is a naturally occurring process but is rarely complete, and failure becomes more extensive with increased disease duration [2].

Decreased oligodendrocyte precursor cell (OPC) differentiation may play a role in remyelination failure [3]. The promotion of remyelination through increased OPC differentiation may provide functional recovery and prevent or reduce irreversible disability accumulation. OPC differentiation is controlled by a range of molecules and their receptors, including constitutively active histamine H₃ receptors expressed by neurons and OPCs [4]. The importance of constitutive H3R activity in OPC differentiation is demonstrated by the ability of inverse agonists, which inhibit both ligand-independent and ligand-dependent receptor activation but not neutral antagonists of H3R, to promote OPC differentiation [4, 5].

GSK239512 is a potent, selective, orally bioavailable and brain penetrant H₃ receptor antagonist/inverse agonist, originally investigated for the treatment of cognitive impairment in Alzheimer’s disease (AD) and schizophrenia [6‐8]. Previous clinical studies have indicated that GSK239512 is well tolerated as a monotherapy when titrated up to 80 µg, resulting in >90% H₃ receptor occupancy in the brain at trough plasma concentrations [6, 7, 9]. GSK239512 promotes OPC differentiation in vitro and enhances remyelination in the cuprizone mouse model of remyelination [4].

Magnetisation transfer ratio (MTR) imaging on magnetic resonance imaging (MRI) can be used to assess myelin content in vivo. MTR changes have been demonstrated to correlate with myelin content in animal models [10‐13] and humans [14‐17]. The aim of this study was to determine the efficacy of GSK239512 in terms of lesion remyelination in patients with RRMS, using changes in MTR as a marker of remyelination.

Of the 153 patients screened across 35 sites between 1 February and 19 August, 2013, 131 were randomised at 34 sites (65 to GSK239512; 66 to placebo). In total, 22 patients failed screening: 20 did not meet inclusion or exclusion criteria and two withdrew consent. All randomised patients were included in the ITT population and completed the titration period. A total of 114 patients, 51 (78%) in the GSK239512 and 63 (95%) in the placebo groups completed both the treatment and follow-up phases. Reasons for withdrawal are shown in Fig. 1.

In total, 55 (85%) and 65 (98%) patients in the GSK239512 and placebo groups, respectively, were able to titrate to 80 µg by the end of the titration period. During the maintenance period, five patients down-titrated from 80 to 40 µg in the GSK239512 group. Overall, patient demographics and baseline characteristics in both treatment groups were as expected for the target population and were balanced between treatment groups (Table 1). The mean and median number of days since relapse was greater in the GSK239512 compared with the placebo group.

Overall, 92 lesions in 27 patients in the GSK239512 group and 97 lesions in 28 patients in the placebo group contributed to the primary analysis of GdE lesions. For Delta-MTR lesions, 69 lesions in 24 patients in the GSK239512 group and 77 lesions in 29 patients in the placebo group contributed to the primary analysis. The analysis excluded all MRI data 70 days pre- and post-reference lesion, and indicated that effect sizes for the change in GdE and Delta–MTR lesions were 0.344 (90% CI 0.018, 0.671) and 0.243 (90% CI −0.112, 0.598), respectively. The adjusted mean change in normalised MTR (difference between normal WM and GM) for GdE lesions was 0.149 (90% CI 0.008, 0.289; Fig. 2a) and Delta-MTR lesions was 0.105 (90% CI −0.049, 0.257; Fig. 2b) favouring GSK239512 over placebo (Table 2). These values are in calibrated MTR units [30] and are interpretable as improved MTR recovery corresponding to 15% (GdE) and 10% (Delta-MTR) of the difference between normal GM and WM. In the Bayesian analysis, estimated probabilities were 0.955 and 0.877 that the effect size for GSK239512 relative to placebo was greater than 0, for the change in post- versus pre-lesion for GdE and Delta-MTR lesions, respectively (Online Resource 2).

At Week 48, there was an increase in lesion development observed trending in favour of placebo (Table 3); post hoc investigations identified an interaction between treatment group and GdE lesion presence on the screening MRI, indicating that patients in the small subgroup with no GdE lesions at the screening MRI had more than double the number of new lesions, on average, during the treatment period on GSK239512 compared with placebo (Online Resource 3). Over the duration of the study, observed changes in brain volume were small. At Week 48, no difference in evolution of whole brain, WM or GM volume was observed between treatment groups (Online Resource 4).

A total of 50 relapses (24 with GSK239512 and 26 with placebo were reported in 38 patients [18 with GSK239512 and 20 with placebo]) during the treatment phase. No difference in relapse rates between treatment groups was observed (GSK239512: 0.417 and placebo: 0.400 relapses per 48 weeks). No meaningful changes from baseline were observed on EDSS and CogState battery throughout the treatment phase and at Week 48 (Table 4). Changes from baseline in MSQoL at Week 48 are shown in Online Resource 5.

Overall, during the treatment phase the incidence of AEs was similar between GSK239512 (74%) and placebo (76%) and AEs were generally of mild and moderate intensity (28 and 42% for GSK239512 and 29 and 42% for placebo, respectively). However, during the titration period the incidence of AEs was higher with GSK239512 (52%) than with placebo (35%), particularly the incidence of insomnia (31 versus 9%, respectively), whereas in the maintenance period the overall incidence of AEs was similar (66 and 67%, respectively). The incidence of AEs judged by investigators as treatment-related was higher with GSK239512 versus placebo during the titration period (35 and 9%, respectively) than the maintenance period (11 and 3%, respectively). As expected, the most commonly reported AE during the treatment phase was insomnia, reported by 22 (34%) and seven (11%) patients in the GSK239512 and placebo groups, respectively; the 95% CI for the relative risk [3.19 (95% CI 1.57, 6.95)] excluded 1, suggesting a real difference. Other incidences of common AEs are shown in Table 5. Three SAEs were reported in two patients [oral papilloma and thrombophlebitis of the leg (n = 1) and thymoma (n = 1)] in the GSK239512 group and one SAE [nephrolithiasis (n = 1)] in the placebo group, during the maintenance period, with no SAEs reported during the titration period. In total, seven patients withdrew from the study due to AEs during the treatment phase, all in the GSK239512 group [insomnia (n = 2), thrombophlebitis (n = 1), cough (n = 1)], elevated alanine aminotransferase (ALT) and aspartate aminotransferase (AST; n = 1), thymoma (n = 1) and, axillary candidiasis, dry mouth, toxic skin eruption and xerosis (n = 1)]. No deaths were reported. Mean haematology, chemistry, urinalysis and vital sign values remained within normal ranges throughout the duration of the study, with the exception of one patient in the GSK239512 group, on a background of intramuscular interferon-β1a treatment, who had moderate level ALT and AST elevations. These AEs were judged by investigators as not treatment-related and this patient was withdrawn from the study.

Minimum trough concentrations of GSK239512 were approximately 10–25 pg/mL for the 80 µg dose and were consistent over the study duration. No relationship between trough concentrations of GSK239512 and the primary endpoints were observed.

In this phase II, randomised, parallel-group, placebo-controlled study, GSK239512 at a maximum dose of 80 µg per day up-titrated over a 4- to 5-week period and maintained for 43–44 weeks had a positive effect on change in MTR, a marker of lesion remyelination in patients with RRMS.

Patient demographics and characteristics were as expected for patients with RRMS who met the defined eligibility criteria. While patient relapse data at screening may be suggestive of a more active population in the GSK239512 treatment group, the current analysis was based on lesion activity during the time period of the study. The on-study disease activity was broadly comparable between treatment groups with a similar mean relapse rate, proportion of patients with relapse and primary endpoint lesion activity. Therefore, the difference in baseline relapse activity did not appear to have any implications for the outcomes of this trial.

The primary MTR analysis focused on the comparison between pre-lesion and recovery values at times when these variables were relatively stable. GdE and Delta-MTR lesions that formed following the baseline MRI were identified using independent procedures, based on images with different contrast mechanisms. While GdE lesions are short-lived and sensitive to blood–brain barrier breakdown [31, 32], Delta-MTR lesions are persistent and appear to better identify tissue that has demyelinated [24]. Although their occurrence does not always co-localize, MTR recovery in either lesion type was expected to show the same response to treatment. This is indeed what we observed, with a small positive effect of GSK239512 versus placebo on lesion MTR being observed with effect sizes of 0.344 in GdE and 0.243 in Delta-MTR lesions with GSK239512 over placebo indicating a relative increase in remyelination, with high posterior probabilities of the effect size being >0 (0.955 in GdE and 0.877 in Delta-MTR lesions, respectively). Therefore, the MTR methodology employed in this study for assessing remyelination represents a promising tool for detecting changes in lesion myelin content in RRMS. However, in our study, the effect size was smaller than the target of 0.5. Additionally, the effect size in this trial was greater for Gd lesions compared with newly detected MTR lesions (Delta-MTR lesions) and may be associated with the greater number of GdE lesions included in the analysis due to relatively high-frequency (approximately monthly) scanning [24]. The use of MTR as a marker of remyelination has been previously validated, demonstrating that MTR values decrease with demyelination and increase with remyelination [10‐17]. Studies on post-mortem brains from patients with MS have shown a strong association between MTR and histopathologically measured myelin content. The MTR of remyelinated lesions differs from both normal-appearing WM and demyelinated lesions [14, 16], and there is a significant correlation between myelin content and MTR in both lesions and the normal-appearing WM [14]. The methodology used to quantify changes in MTR over time associated with lesion remyelination has been previously reported [24].

Secondary MRI assessments did not demonstrate a positive impact of GSK239512 treatment that corresponded with the observed improvement in remyelination demonstrated by the MTR assessment. While unexpected from the pharmacologic action of the study drug, a small relative difference in favour of placebo was found. The magnitude of effect is unlikely to be clinically relevant, as analysis of pooled data from clinical trials has shown that an increase in relapse frequency is predicted only by larger differences in active lesion detection [33]. The absence of changes in secondary assessments, particularly MTR in chronic T2-weighted lesions may indicate the need for further development of additional remyelination marker endpoints, capable of demonstrating changes over time associated with improvement in remyelination [34, 35].

As a H₃ receptor antagonist/inverse agonist, GSK239512 was not expected to improve or enhance the inflammatory disease process. Importantly, there was no indication that GSK239512 had an exacerbating effect, with similar relapse rates and EDSS scores observed in both treatment groups. No meaningful changes in brain volumes were observed; this was not unexpected given the relatively small changes in remyelination of lesions and extended time period required to observe changes in brain volume measurements in a sample of this size. Additionally, the patient population assessed had modest physical and cognitive impairment at baseline. While memory scores on the CogState battery improved slightly in both treatment groups this may be an artefact of learning or exercise effect.

Safety findings were consistent with previous studies of GSK239512 in AD and schizophrenia [6‐8]. GSK239512 was well tolerated by the majority of patients, with 85% of patients reaching the 80 µg dose. Patients treated with GSK239512 reported more CNS-related AEs such as insomnia and nightmare than with placebo and sleep-related AEs were found to be the most common side effect of GSK239512; this is unsurprising for a RRMS population and the treatment mechanism of action, and is consistent with previous trials of GSK239512 in patients with AD [6, 7]. In total, seven patients withdrew due to AEs, all in the GSK239512 group and AE tolerability may have been a contributing factor to the differential withdrawal rate between the placebo and GSK239512 groups. Several of these were judged to be potentially treatment-related including liver function test elevation and toxic skin eruption and may be due to an interaction between GSK239512 and concomitant interferon-1βa therapy and should therefore be considered in future study designs. Trough plasma concentrations of GSK239512 at the 80 µg dose are consistent with >90% H₃ receptor occupancy in the brain, as predicted from a previous positron emission topography study conducted in healthy human subjects [9].

One limitation of this trial was that not all patients contributed at least one lesion to the analysis as expected, making the results harder to generalise. Further studies could assess the effect of potential remyelinating agents in a wider patient population, including patients without lesions to further extend generalisability. Additionally, there was no clinical correlate of improved remyelination. Clinical outcome measures used generally require much larger sample sizes or longer study durations (2–3 years) to demonstrate clinically relevant changes. Future studies should consider evaluating the use of shorter-term clinical endpoints, include more patients and define a patient population with the capability of demonstrating a change. Additionally, targeted clinical deficits associated with specific demyelinating lesions might be more sensitive and could be reflected in additional surrogate endpoints [34, 35].

In summary, GSK239512 treatment resulted in a positive effect on lesion remyelination, detected by GdE and Delta-MTR lesion assessments, validating this approach for multicentre clinical trials. GSK239512 was found to have an acceptable safety profile in patients with RRMS. However, the positive effect on lesion remyelination did not translate into observed benefits on conventional MRI or clinical assessments. The development of clinical endpoints more directly linked to changes in myelination, inclusion of a population with more disease activity and/or a longer duration to match the clinical assessments used in the current trial are required to further explore the potential impact of an H₃ receptor antagonist/inverse agonist on remyelination and ultimately disease progression in RRMS.