Long-term data
Safety studies
Comment: Dr. Mark S. Freedman, Director, Multiple Sclerosis Research Unit, The Ottawa Hospital, Ottawa, Canada
The 31st congress of the European Committee for Treatment and Research in MS (ECTRIMS) featured new research on current therapies for multiple sclerosis. In Canada, two oral therapies – teriflunomide and dimethyl fumarate (DMF) – are routinely used as first-line agents. The following summarizes key data for these two therapies presented at ECTRIMS.
Efficacy analyses
Teriflunomide: The efficacy of teriflunomide in relapsing MS was demonstrated in the phase III TEMSO and TOWER trials (O’Connor et al. N Engl J Med 2011;365:1293-1303; Confavreux et al. Lancet Neurol 2014;13:247-256). A new post-hoc analysis of TEMSO examined the proportion of patients who achieved no evidence of disease activity (NEDA), defined as no relapses, no new MRI activity and no disability progression. Overall, 28.1% of patients receiving teriflunomide 14 mg achieved NEDA compared to 14.3% with placebo (odds ratio 2.34); NEDA was calculated from months 6-24, when treatment was fully effective (Chan et al. ECTRIMS 2015; abstract 1037).
In the original TEMSO and TOWER cohorts, 5.4% of randomized patients had progressive MS. Data were re-analysed to determine the annualized relapse rate (ARR) in those with relapsing-remitting MS (Comi et al. ECTRIMS 2015; abstract P1048). ARR was reduced 35.4% with teriflunomide 14 mg versus placebo, which was similar to what was reported for the full dataset of patients treated with the higher dose (35.5-36.3% reduction). The proportion of patients with 12-week confirmed disability progression was reduced 30.4% with teriflunomide 14 mg versus placebo, which was comparable to the results for the full cohort (30-32% reduction). The rate of disability progression was 44.7% lower in the subgroup (n=124) with progressive MS; results were not significant due to the small sample size.
A separate analysis examined disability worsening in patients with and without relapses in the pooled TEMSO/TOWER population (O’Connor et al. ECTRIMS 2015; abstract P1031). Overall, 87% of patients who were relapse-free had no disability worsening. For the 22% of patients with confirmed disability progression, 84.5% experienced worsening disability within 30 days after experiencing a relapse, suggesting that short-term disability is largely attributable to relapse-related residual deficits.
The two phase III studies of teriflunomide demonstrated a consistent effect of treatment on disability progression, but there was only a non-significant 25% reduction in brain volume loss in TEMSO (MRI was not used in TOWER); the method used was an MRI image analysis package. Brain volume was re-analysed with SIENA (Structural Image Evaluation, Using Normalization of Atrophy), which compares images at two time points to estimate brain volume change. With this method, median brain volume change from baseline was 36.5% lower with teriflunomide 14 mg/day group compared to placebo at one year, and 30.6% lower at two years (Radue et al. ECTRIMS 2015; abstract LB229). The rate of brain volume loss was also significantly lower with teriflunomide 7 mg/day compared to placebo at the two time points (34.4%; 27.6%).
DMF: The efficacy of DMF in relapsing MS was shown in the phase III DEFINE and CONFIRM trials (Gold et al. N Engl J Med 2012;367:1098-1107; Fox et al. N Engl J Med 2012;367:1087-1097); patients were then eligible to enter the ENDORSE long-term extension. An integrated analysis of DEFINE/CONFIRM data examined the subgroup of patients with less baseline disability (EDSS score < 2) (Gold et al. ECTRIMS 2015; abstract 565). There was a significant 32% reduction in the risk of disability progression in the DMF group compared to placebo, which is similar to the 38% risk reduction reported for all patients in DEFINE.
The BVL rate was calculated for patients in ENDORSE completing six years of treatment with DMF 240 mg BID for whom MRI results were available (Miller et al. ECTRIMS 2015; abstract P1072). The adjusted annualized mean percent brain volume change (PBVC) over six years was -0.34%/year, which is comparable to the -0.27%/year rate recently reported for healthy controls (De Stefano et al. J Neurol Neurosurg Psychiatry 2015; epublished April 22, 2015). However, the data set (n=87) represented only 11% of the original 240 mg BID cohort in DEFINE/CONFIRM.
Long-term data
Teriflunomide: The efficacy of teriflunomide at 4.5 years was evaluated in the extension of the TOWER study (Kappos et al. ECTRIMS 2015; abstract P1099). ARR declined from 0.32 at the end of the two-year core study to 0.257 with teriflunomide 14 mg/day at 2.5 years of the extension. A total of 61.2% of teriflunomide-treated patients were relapse-free (O’Connor et al. ECTRIMS 2015; abstract 555). The mean change in EDSS score was 0.13 for years 2-4.5. The proportion with 12-week confirmed disability progression was 22.4% (Kappos 2015). In comparing patients initially randomized to active therapy versus those switching from placebo to teriflunomide at the end of the core study, there was a benefit to earlier versus later therapy with respect to the proportion of patients who were relapse-free (61.6% vs. 41.1%) and with no disability progression (73.3% vs. 69.8%) (O’Connor 2015).
A post-hoc analysis of TEMSO data reported that the cross-sectional NEDA rate remained >40% in any given year of treatment with teriflunomide 14 mg during the first six years of the extension (Wolinsky et al. ECTRIMS 2015; abstract P1047).
An alternative method of evaluating treatment response is the modified Rio score (Sormani et al. Mult Scler 2013;19:605-612), which categorizes disability risk based on treatment response as Low (score 0; <5 new T2 lesions, no relapses), Intermediate (score 1; <5 new T2 lesions + 1 relapse or >5 new T2 lesions, no relapses) or High (score 2; <5 new T2 lesions + >2 relapses or >5 new T2 lesions + 1 relapse; or score 3; >5 new T2 lesions + >2 relapses). A score of 0 corresponds to a treatment response; and a score of 2-3 is considered a treatment non-response.
A modified Rio score was determined for patients completing one year of TEMSO; patients were reclassified as responders or non-responders six months later according to clinical and MRI outcomes (Sormani et al. ECTRIMS 2015; abstract P1131). Overall, 79.2% of patients receiving teriflunomide 7 mg or 14 mg were responders versus 20.8% of patients in the placebo group. The probability of 12-week sustained disability progression was lower for responders versus non-responders within the first few months of treatment, and the risk of progression was significantly lower at a median 5.6-year follow-up (hazard ratio 2.055).
DMF: In the ENDORSE extension study, the cumulative ARR over six years was 0.164 (Hutchinson et al. ECTRIMS 2015; abstract P543). The median EDSS score was unchanged (2.0) during years 3-5. At year 6, 73% of patients receiving DMF 240 mg BID were free of new/enlarging T2 lesions, and 87% had no Gd+ T1 lesions (Arnold et al. ECTRIMS 2015; abstract 566).
The predictive value of NEDA was analysed by comparing long-term outcomes of patients with or without NEDA at the end of DEFINE/CONFIRM and completing four years of the ENDORSE extension (Havrdova et al. ECTRIMS 2015; abstract P557); the subgroup (n=799) represented 30% of the original ITT population of the two studies (70% of the original MRI cohort). Overall, 171 of 799 patients (21%) achieved NEDA at the end of two years of treatment. The adjusted ARR was significantly lower in the NEDA versus non-NEDA group (0.08 vs. 0.20). The difference in mean EDSS score at six years was also significant in the NEDA versus non-NEDA groups (2.3 vs. 2.6). A similar proportion of patients in ENDORSE remained on therapy (about 70%) whether or not NEDA was achieved.
Safety studies
Teriflunomide: Longer-term safety was assessed in the 2.5-year extension of the TOWER study (Freedman et al. ECTRIMS 2015; abstract EP1460). Cumulative exposure was 443 patient-years for patients receiving teriflunomide 14 mg throughout the study. Most adverse effects were mild or moderate. The most common adverse events were nasopharyngitis (13.2%), headache (9.2%), diarrhea (8.4%), hair thinning (7.6%), back pain (7.1%), and ALT increase (6.3%). In 62% of the instances of elevated ALT, the increase was ≤ 3 times the upper limit of normal (ULN). The incidence of infections (46.1% vs. 38.6%) and serious infections (1.3% vs. 1.2%) was similar with teriflunomide 14 mg and placebo. There was no evidence of an increased risk of malignancy with treatment.
An observational study at nine MS centres in the U.S. examined the course of hair thinning in affected patients (n=38; mean age 52 years, 97% female). Patients were evaluated at the onset of hair thinning and at a six-month follow-up visit; photographs were taken at each visit and evaluated by a dermatologist. The mean time from treatment initiation to onset of hair thinning was 77 days. Healthcare professionals rated the severity of hair thinning as mild (63%) or moderate (34%); one case was severe in a patient with a previous history of drug-induced hair loss. A complete or near-complete resolution or marked improvement in hair thinning was reported by 79% of patients at follow-up. The time to hair regrowth was not reported. Overall, 1 of 38 patients permanently discontinued teriflunomide and one patient interrupted treatment due to hair thinning.
DMF: An emerging safety issue is progressive multifocal leukoencephalopathy (PML) following the report of the first case with branded DMF a year ago; three cases have now been reported (October 2015). It has been speculated that PML risk may be associated with persistent lymphopenia (absolute lymphocyte count < 0.5 x 109/L) during DMF therapy, although this has not yet been established.
In DEFINE, WBC < 3.0 x 109/L or ALC < 0.5 x 109/L was seen on at least one occasion in 4% of DMF-treated patients (Gold 2012, supplemental data). An analysis of clinical trial data (phase IIb, DEFINE, CONFIRM; n=2,513) reported that ALC remained within normal limits at all time points in 60.1% of DMF-treated patients (Fox et al. ECTRIMS 2015; abstract P606). Overall, 7.2% experienced severe lymphopenia (< 0.5 x 109/L) at any time; 2.5% experienced severe lymphopenia persisting for >6 months. Data on lymphocyte recovery were available for only 14 of 53 patients (26%) with persistent severe lymphopenia. In this subgroup, lymphocyte counts reportedly increased in 12 of 14 patients following DMF discontinuation.
Three real-world studies of lymphocyte counts during DMF treatment were reported at ECTRIMS. At the University of Washington, Seattle, the incidence of severe lymphopenia among 213 patients receiving DMF was 13% (Romba et al. ECTRIMS 2015; abstract P1130). Time to ALC >0.5 x 109/L was six months after DMF discontinuation. Risk factors for lymphopenia were older age, higher EDSS score and lower lymphocyte count prior to DMF initiation. Similarly, age and disease duration were risk factors for severe lymphopenia in a retrospective analysis of 335 patients treated with DMF at the Portland, Oregon, MS centre (Smoot et al. ECTRIMS 2015; abstract P610). The incidence of severe lymphopenia on at least one occasion was 9.6%. A retrospective study in Scotland reported that 11 of 174 patients (6.3%) receiving DMF for >6 months had severe lymphopenia (MacDougall et al. ECTRIMS 2015; abstract P624). Low lymphocyte counts persisted in two patients at five-month follow-up despite DMF discontinuation.
Comment
Dr. Mark S. Freedman: ECTRIMS brought a wealth of new scientific insight about MS, important follow-up on new and older treatments, new concepts and goals of treatment response, and strategies to monitor and mitigate safety issues concerning the disease-modifying medications.
All phase III studies in RRMS are two years in duration, but the hope is that treatment of patients goes for much longer. Studies often do away with the placebo or comparator arm after two years, but there is still some value in following study patients into later years, looking for signs of better treatment response or new safety issues that did not arise during early exposure. A popular new measure of efficacy is NEDA – or complete lack of evidence of ongoing measurable disease. Given that many patients, even in the placebo arm, remain with NEDA after three years, what is the value of measuring this outcome over the usual three main outcome measures of relapse, MRI activity and EDSS progression? At least two points come to mind. If patients have NEDA due to the treatment, they are spared all the possible indicators of disease, whereas simply having fewer relapses need not be accompanied by less MRI activity or progression. Secondly, if more patients have NEDA after one or two years, it may well be a good prognostic for continued treatment response. This is usually expressed as the increased odds of having NEDA compared to the placebo or comparator. Such a predictive value was seen in the pooled studies for DMF, however, the data represented only a small portion (30%) of the original treatment group due to patient drop-outs and MRIs performed in less than 40% of all patients.
Although we measure relapses as the primary outcome of today’s studies, the real hope is that the attacks averted are meaningful and would have likely contributed to progression. That is why it is important not just to count fewer relapses, but to look at the quality or severity of the attacks that are reduced. Such an analysis of the teriflunomide studies indicated a relationship between having attacks and EDSS progression. Most of the relapse-free patients were also disability-free, and most of the patients showing EDSS progression did so within a month of a relapse. Progression changes also must be meaningful. In an analysis of the pooled DMF datasets, there was a special look at patients with EDSS < 2 to demonstrate the effect of DMF in slowing progression. However, would progressions in the range of <2 be at all meaningful? This point reminds us of the ordinal EDSS scale, in which changes in the lower end are much less meaningful than changes in the higher range above 3.
Longer term follow-up of teriflunomide patients has not revealed any safety issues not seen in the first couple of years. That is in contrast to the findings now of PML in association with DMF or fingolimod. There is no clear indicator as to which patients on either of those two therapies are more at risk – whether the antibody status to JC virus is useful, or whether lymphopenia – seen in the association of PML with the use of fumaric acid in psoriasis – will also be an indicator of PML risk with DMF. In the case of fingolimod, lymphopenia is expected, although this does not indicate true lymphopenia as the cells are sequestered in lymph nodes. However, with DMF, lymphopenia is not expected and may represent a toxicity predisposing to PML.
Finally, a word about measuring brain volumes and interpreting such changes. In the case of DMF, the studies were hampered by MRIs being performed in only a subset of the entire patient groups, about 40% overall, raising the question as to whether these patients truly represent how the entire cohort would do. Though rarely examined, it would be important to show that patients saved of atrophy were also spared EDSS progression and vice versa, but such correlations are rarely evaluated. In the case of teriflunomide, the question was raised as to how this medication – being the only approved oral agent to demonstrate a significant effect on reducing EDSS progression – did not seem to have an effect on slowing brain atrophy. It turns out that unless all measurements of brain volume or its corollary of brain atrophy are done in the same way, then values are simply not comparable between studies. A re-analysis of the teriflunomide data using the SIENA methodology revealed that teriflunomide did indeed have a significant effect on reducing brain atrophy.