Delay From Treatment Start to Full Effect of Immunotherapies for Multiple Sclerosis

Izanne Roos; Emmanuelle Leray; Federico Frascoli; Romain Casey; J. William L. Brown; Dana Horakova; Eva K. Havrdova; Maria Trojano; Francesco Patti; Guillermo Izquierdo; Sara Eichau; Marco Onofrj; Alessandra Lugaresi; Alexandre Prat; Marc Girard; Pierre Grammond; Patrizia Sola; Diana Ferraro; Serkan Ozakbas; Roberto Bergamaschi; Maria José Sá; Elisabetta Cartechini; Cavit Boz; Franco Granella; Raymond Hupperts; Murat Terzi; Jeannette Lechner-Scott; Daniele Spitaleri; Vincent Van Pesch; Aysun Soysal; Javier Olascoaga; Julie Prevost; Eduardo Aguera-Morales; Mark Slee; Tunde Csepany; Recai Turkoglu; Youssef Sidhom; Riadh Gouider; Bart Van Wijmeersch; Pamela McCombe; Richard Macdonell; Alasdair Coles; Charles B. Malpas; Helmut Butzkueven; Sandra Vukusic; Tomas Kalincik


Brain. 2020;143(9):2742-2756. 

In This Article


Patients and Follow-up

Of 125 421 patients (60 662 MSBase, 64 759 OFSEP) assessed for study inclusion, the numbers of patients eligible for analysis of time to full clinically manifest effect of therapy were 9147 (5391 MSBase, 3756 OFSEP) for the relapse outcome and 3581 (2339 MSBase, 1242 OFSEP) for the progression-of-disability events (Supplementary Figure 2). The number of patients per contributing centre is shown in Supplementary Table 2 and their clinical and demographic characteristics are shown in Table 1. Apart from a longer multiple sclerosis duration, the clinicodemographic details of the included population were similar to patients who received multiple sclerosis therapies but were excluded from the study (Supplementary Table 3). Characteristics of the study population stratified by therapy are available in Supplementary Table 4.

Relapse Cohort. After exclusion of insufficiently represented therapies, 11 DMTs with 11 180 treatment epochs in 9013 patients (5325 MSBase, 3688 OFSEP) were analysed (Table 1). Overall, 75% of patients were female, with a mean age at treatment initiation of 39.9 years, a median disease duration at baseline of 10.4 years and an ARR of 0.47 (0.27–0.74). The median pre-baseline follow-up duration was 6.6 years. Twenty-three per cent of patients were treatment naïve at baseline and in 57% the preceding therapy was an injectable DMT. Lack of efficacy was the most commonly reported reason for discontinuation of the preceding DMT. Population characteristics were similar between registries.

Disability Progression Cohort. From 3531 patients, 4088 (2682 MSBase, 1406 OFSEP) treatment epochs were obtained and seven sufficiently represented DMTs (Table 1). The median baseline EDSS was 3 (2–4.5) and 13.7% of the cohort had progressive multiple sclerosis. Despite the requirement of 1-year treatment persistence the index DMT was continued for a median of 4.6 years (range 2.4–5 years). Patients in the MSBase cohort had a lower median EDSS (3 versus 3.5) and a higher proportion of pre-baseline time on treatment (0.59 versus 0.37) than those in OFSEP.

Proof of Principle: The Method to Identify Therapeutic Lag

In the discovery analysis, data from MSBase were used to calculate the time to full clinically manifest effect of treatment for relapses in nine adequately represented therapies (Table 2). Tr was identified in all DMTs with more than 200 relapse events and ranged between 12.6 and 27.5 weeks. In the validation analysis (data from OFSEP) Tr was identified for four of eight DMTs analysed (natalizumab, teriflunomide, interferon beta-1b, subcutaneous interferon beta-1a and intramuscular interferon beta-1a) and ranged between 12.3 and 23.1 weeks; an insufficient number of events were available to find the first local minimum of the first derivative for fingolimod, dimethyl fumarate, glatiramer acetate and interferon beta-1b. The mean Tr estimated in MSBase and OFSEP were similar. The bootstrapped mean (95% CI) differences in Tr (in weeks) between the registries were 1 (−5.6, 10.7) for teriflunomide, −8.8 (−20, 5.2) for subcutaneous interferon beta-1a, −2.6 (−20, 3.2) for intramuscular interferon beta-1a and 5.2 (−4.5, 22.4) for natalizumab.

On exploration of the effects of the shape of the density curve on Tr, no correlation between the height of the peak of the density curve and Tr was observed (Supplementary Figure 3). Results from a sequential analysis, analysing the number of relapses required for stable estimates of Tr, are shown in Supplementary Figure 4. The minimum number of events required to estimate a consistent and stable Tr varied in response to variation in the shape of the relapse density curves. For all DMTs, the variability of Tr estimates reduced with increasing number of relapses sampled. Whilst Tr reached stability with fewer than 1000 relapses for most therapies, a minimum of 2000 relapse events were required for stability of Tr on natalizumab, fingolimod and glatiramer acetate.

Therapeutic Lag for Relapses

Time to Treatment Effect. In the combined MSBase-OFSEP cohort of 9013 patients with 11 180 treatment epochs, 23 424 relapses were recorded. For all 11 studied therapies, an increase in relapse density preceded the initiation of the index DMT, reflecting the events leading to commencement of new treatments, and was followed by a subsequent decline in relapse occurrence (Figure 2). Initial decline in relapse density was observed even prior to the start of index therapy; this artefact results from delay between a relapse and commencement of next therapy and the fact that all events within 30 days from a prior relapse constitute a single relapse. The calculated time to full clinically manifest treatment effect on relapses (Tr) is displayed in Table 3 and Figure 2. Mean time to treatment effect ranged between 9.4 and 19.8 weeks for all treatments other than dimethyl fumarate, which showed Tr of 30.2 weeks (95% CI 26.6–33.7). A bimodal distribution of Tr was present for fingolimod, with 85% of estimates and the highest density probability occurring at 12.7 weeks. For all treatments, the bootstrapped Tr estimate with the highest density probability mirrored the point identified from the entire available population. There were insufficient number of relapses on rituximab (304 relapses in 101 treatment epochs) to identify Tr. Despite only 220 relapses on alemtuzumab, Tr was identified at a mean of 16 weeks (95% CI 14.6–17.3) with satisfactory stability in the sequential analysis (Supplementary Figure 4).

Figure 2.

Duration of therapeutic lag for relapses. Density curve indicates the ARR and duration of therapeutic lag for relapses. The first post-treatment local minimum, representing the point of stabilization of treatment effect, is indicated in purple (Tr). The last stable point prior to start of the index therapy is indicated in orange. Two black lines indicate the upper and lower bounds of 95% CIs. Density curves from Monte Carlo simulations are indicated in grey.

A sensitivity analysis restricting relapses to those necessitating treatment with corticosteroids or resulting in a change in EDSS showed consistent results for all adequately represented DMTs (Supplementary Table 3).

Magnitude of Treatment Effect. For each therapy, ARRs were compared in the pre- and on-treatment periods (Figure 3). With the exception of intramuscular interferon beta-1a, all therapies were associated with a drop-in relapse activity on-treatment versus pre-baseline. This suggests that over the short term, the chosen therapies led to the desired improved control of disease activity.

Figure 3.

ARRs in the pretreatment and on-treatment periods. The pretreatment ARR was calculated from the start of the treatment epoch to the last point of stabilization prior to treatment start, dividing relapse density by the cumulative follow-up time prior to that point. The on-treatment ARR was calculated from Tr to the end of treatment epoch, dividing relapse density by the respective cumulative follow-up time.

Therapeutic Lag for Confirmed Progression-of-disability Events. In the 4088 included treatment epochs from 3531 patients, 2563 disability progression events were identified across seven DMTs. On inspection of the progression-of-disability density curves, increase in the likelihood of progression-of disability preceded commencement of the index DMT (Figure 4). Similar to relapses, progression event rates in the pre-baseline period were highest for patients commenced on higher efficacy therapies, natalizumab and mitoxantrone. The occurrence of progression-events after initial stabilization at Td resumed to increase for most studied DMTs. However, for mitoxantrone and interferon beta-1b, the number of progression-of-disability events continued to decline throughout the 5-year follow-up period. The occurrence of relapse independent progression-of-disability events reduced for ~1–2 years after commencement of therapy and resumed to increase thereafter for most DMTs (Supplementary Figure 5).

Figure 4.

Duration of therapeutic lag for progression of disability. Density curve representing the number of progression events per 100 patient-years and duration of therapeutic lag for progression of disability. The first post-treatment local minimum, representing the point of stabilization of treatment effect, is indicated in purple (Td). Two vertical black lines indicate the upper and lower bounds of 95% CIs. Density curves from Monte Carlo simulations are indicated in grey.

The calculated duration of therapeutic lag on progression-of-disability events (Td) across therapies is shown in Table 4 and Figure 4. Td for the seven sufficiently represented therapies ranged between 30 and 52 weeks, with the exception of intramuscular interferon beta-1a, for which the mean Td was estimated at 70.4 weeks (95% CI 59.8–81). An insufficient number of relapse independent progression-of-disability events (TPIRA) were present to calculate the duration of therapeutic lag for any DMT.