نقش تری‌فلونوماید در درمان مالتیپل اسکلروزیس
چکیده
پیشینه
این یک نسخه به‌روز شده از مرور کاکرین با عنوان «نقش تری‌فلونوماید (teriflunomide) در درمان مالتیپل اسکلروزیس» است (برای نخستین‌بار در کتابخانه کاکرین (The Cochrane Library )، سال 2012، شماره 12، منتشر شد).
مالتیپل اسکلروزیس (multiple sclerosis; MS) یک بیماری مزمن سیستم عصبی مرکزی با واسطه ایمنی است. مشخصه بالینی آن، عود مکرر یا پیشرفت بیماری، یا هر دو، است که اغلب منجر به ناتوانی شدید عصبی و کاهش جدی در سطح کیفیت زندگی می‌شود. هدف از درمان‌های اصلاح‌کننده بیماری (disease‐modifying therapies; DMTs) برای MS، پیشگیری از وقوع عود و پیشرفت ناتوانی است. تری‌فلونوماید یک مهار کننده سنتز پیریمیدین (pyrimidine synthesis inhibitor) است که توسط سازمان غذا و داروی آمریکا (FDA) و آژانس دارویی اروپا (EMA) به عنوان یک DMT برای بزرگسالان مبتلا به MS عود کننده‐بهبود یابنده (relapsing‐remitting MS; RRMS) تایید شده است.
اهداف
ارزیابی اثربخشی و بی‌خطری (safety) مطلق و مقایسه‌ای تری‌فلونوماید به عنوان تک درمانی (monotherapy) یا درمان ترکیبی در برابر دارونما (placebo) یا دیگر داروهای اصلاح‌کننده بیماری (DMDs) (اینترفرون بتا (interferon beta; IFNβ)، گلاتیرامر استات (glatiramer acetate)، ناتالیزوماب (natalizumab)، میتوکسانترون (mitoxantrone)، فینگولیمود (fingolimod)، دی‌متیل فومارات (dimethyl fumarate)، آلمتوزوماب (alemtuzumab)) برای تعدیل سیر بیماری در افراد مبتلا به MS.
روش‌های جست‌وجو
پایگاه ثبت تخصصی کارآزمایی‌های گروه مالتیپل اسکلروزیس و بیماری‌های نادر دستگاه سیستم عصبی مرکزی (CNS) در کاکرین را جست‌وجو کردیم (30 سپتامبر 2015). فهرست منابع مرورهای منتشر شده و مقالات بازیابی شده را بررسی کرده و گزارش‌های ارائه شده (2004 تا سپتامبر 2015) را در انجمن‌های MS در اروپا و آمریکا جست‌وجو کردیم. هم‌چنین با محققین حاضر در کارآزمایی‌های تری‌فلونوماید و شرکت داروسازی Sanofi‐Aventis ارتباط برقرار کردیم.
معیارهای انتخاب
کارآزمایی‌های بالینی تصادفی‌سازی شده، کنترل شده و گروه موازی (parallel‐group) را با طول دوره پیگیری یک سال یا بیشتر در این مرور گنجاندیم که به ارزیابی تری‌فلونوماید، به عنوان تک درمانی یا درمان ترکیبی، در برابر دارونما یا دیگر DMDهای تایید شده، بدون محدودیت در دوز، دفعات تجویز و مدت زمان درمان، برای افراد مبتلا به MS پرداختند.
گردآوری و تجزیه‌وتحلیل داده‌ها
از پروسیجرهای روش‌شناسی (methodology) استاندارد کاکرین استفاده کردیم. دو نویسنده مرور به‌طور مستقل از هم کیفیت کارآزمایی را ارزیابی و داده‌ها را استخراج کردند. اختلاف نظرات مورد بحث قرار گرفت و با اجماع بین نویسندگان مرور حل‌وفصل شد. برای یافتن داده‌های بیشتر یا تایید داده‌ها، با محققین اصلی مطالعات وارد شده تماس گرفتیم.
نتایج اصلی
پنج مطالعه شامل 3231 نفر، اثربخشی و بی‌خطری تری‌فلونوماید را در دوز 7 میلی‌گرم و 14 میلی‌گرم، به تنهایی یا همراه با IFNβ کمکی، در برابر دارونما یا IFNβ‐1a در بزرگسالان مبتلا به اشکال عود کننده MS و نمره مقیاس وضعیت ناتوانی گسترده (Expanded Disability Status Scale) کمتر از 5.5 هنگام ورود ارزیابی کردند.
به‌طور کلی، ناهمگونی‌های بالینی آشکار به دلیل تنوع در طرح‌های مطالعه یا مداخلات و ناهمگونی‌های روش‌شناسی (methodology) در طول مطالعات وجود داشت. همه مطالعات دارای خطر بالای سوگیری تشخیص (detection bias) برای ارزیابی عود و خطر بالای سوگیری (bias) به دلیل تضاد منافع بودند. از میان آنها، سه مطالعه علاوه بر این، به دلیل نرخ بالای ترک مطالعه دارای خطر بالای سوگیری ریزش نمونه (attrition bias) و دو مطالعه نیز دارای خطر نامشخص سوگیری ریزش نمونه بودند. مطالعات درمان ترکیبی با IFNβ (650 شرکت‌کننده) و مطالعه با IFNβ‐1a به عنوان کنترل (324 شرکت‌کننده) نیز به دلیل نمونه محدود، خطر بالایی برای سوگیری عملکرد (performance bias) و فقدان توان آزمون داشتند.
دو مطالعه مزیت و بی‌خطری تری‌فلونوماید را به عنوان تک درمانی در برابر دارونما طی یک دوره یک سال (1169 شرکت‌کننده) یا دو سال (1088 شرکت‌کننده) ارزیابی کردند. متاآنالیز انجام نشد. تجویز تری‌فلونوماید با دوز 7 میلی‌گرم/روز یا 14 میلی‌گرم/روز به عنوان تک درمانی در مقایسه با دارونما، موجب کاهش تعداد شرکت‌کنندگان با حداقل یک عود در طول یک سال (خطر نسبی (RR): 0.72؛ 95% فاصله اطمینان (CI): 0.59 تا 0.87؛ P value = 0.001 با 7 میلی‌گرم/روز و RR: 0.60؛ 95% CI؛ 0.48 تا 0.75؛ P value < 0.00001 با 14 میلی‌گرم/روز) یا دو سال (RR: 0.85؛ 95% CI؛ 0.74 تا 0.98؛ P value = 0.03 با 7 میلی‌گرم/روز و RR: 0.80؛ 95% CI؛ 0.69 تا 0.93؛ P value = 0.004 با 14 روز) شد. فقط تری‌فلونوماید با دوز 14 میلی‌گرم/روز تعداد شرکت‌کنندگان با پیشرفت ناتوانی را طی یک سال (RR: 0.55؛ 95% CI؛ 0.36 تا 0.84؛ P value = 0.006) یا دو سال (RR: 0.74؛ 95% CI؛ 0.56 تا 0.96؛ P value = 0.02) کاهش داد. هنگام در نظر گرفتن تاثیر ترک مطالعه، آنالیزهای سناریوی مورد احتمالی (likely‐case scenario analyses) هم‌چنان در کاهش تعداد شرکت‌کنندگان با حداقل یک بار عود، مزیتی را نشان می‌دهند، اما نه برای تعداد شرکت‌کنندگان با پیشرفت ناتوانی. هر دو دوز هم‌چنین نرخ عود سالانه و تعداد ضایعات T1‐weighted و تقویت شده با گادولینیوم را طی دو سال کاهش دادند. کیفیت شواهد برای پیامدهای عود در یک سال یا دو سال در سطح پائین بود، در حالی که برای پیشرفت ناتوانی در یک سال یا دو سال در سطح بسیار پائین قرار داشت.
تری‌فلونوماید با دوز 14 میلی‌گرم/روز در مقایسه با IFNβ‐1a، اثربخشی مشابه با IFNβ‐1a در کاهش نسبت شرکت‌کنندگان با حداقل یک عود طی یک سال داشت، در حالی که تری‌فلونوماید با دوز 7 میلی‌گرم/روز تاثیر کمتری نسبت به IFNβ‐1a نشان داد (RR: 1.52؛ 95% CI؛ 0.87 تا 2.67؛ P value = 0.14؛ 215 شرکت‌کننده با 14 میلی‌گرم/روز و RR: 2.74؛ 95% CI؛ 1.66 تا 4.53؛ P value < 0.0001؛ 213 شرکت‌کننده با 7 میلی‌گرم/روز). با این حال، سطح کیفیت شواهد بسیار پائین بود.
از نظر مشخصات بی‌خطری، شایع‌ترین عوارض جانبی مرتبط با تری‌فلونوماید عبارت بودند از اسهال، تهوع، نازک شدن مو، افزایش آلانین آمینوترانسفراز، نوتروپنی و لنفوپنی. این عوارض جانبی، تاثیراتی وابسته به دوز داشته و به ندرت منجر به قطع درمان شدند.
نتیجه‌گیری‌های نویسندگان
شواهدی با کیفیت پائین وجود دارد که نشان می‌دهد تری‌فلونوماید با دوز 7 میلی‌گرم/روز یا 14 میلی‌گرم/روز به عنوان تک درمانی در مقایسه با دارونما، هم تعداد شرکت‌کنندگان را با حداقل یک عود و هم نرخ عود سالانه را طی یک سال یا دو سال درمان کاهش می‌دهد. فقط تری‌فلونوماید با دوز 14 میلی‌گرم/روز تعداد شرکت‌کنندگان با پیشرفت ناتوانی را کاهش داد و پیشرفت ناتوانی را در طول یک یا دو سال به تاخیر انداخت، اما کیفیت شواهد بسیار پائین بود. کیفیت داده‌های موجود برای ارزیابی مزیت تری‌فلونوماید به عنوان تک درمانی در برابر IFNβ‐1a یا به عنوان درمان ترکیبی با IFNβ بسیار پائین بود. عوارض جانبی شایع عبارت بودند از اسهال، تهوع، نازک شدن مو، افزایش آلانین آمینوترانسفراز، نوتروپنی و لنفوپنی. این عوارض جانبی عمدتا از نظر شدت خفیف تا متوسط بودند، اما تاثیر وابسته به دوز داشتند. برای ارزیابی مزیت نسبی تری‌فلونوماید بر این پیامدها و بی‌خطری آن در مقایسه با دیگر DMTها، انجام مطالعات جدید با کیفیت بالا و دوره پیگیری طولانی‌تر مورد نیاز است.
PICO
خلاصه به زبان ساده
تری‌فلونوماید سیر بیماری را در افراد مبتلا به مالتیپل اسکلروزیس اصلاح می‌کند
پیشینه
تری‌فلونوماید (teriflunomide) برای نخستین‌بار در آرتریت روماتوئید استفاده شده، و مشخص شد که هم دارای عملکرد آنتی‌پرولیفراتیو (مهار رشد سلولی) و هم ضد التهابی (مقابله با پاسخ موضعی به آسیب سلولی) است. در سال 2012، استفاده از آن برای این ویژگی‌ها توسط سازمان غذا و داروی آمریکا (FDA) برای افراد مبتلا به اشکال عود کننده (با تشدیدهای مکرر نشانه‌های عصبی) مالتیپل اسکلروزیس (multiple sclerosis; MS) و هم‌چنین در سال 2013 توسط آژانس دارویی اروپا تایید شد.
اهداف
ارزیابی اثربخشی و بی‌خطری (safety) دو دوز مختلف از تری‌فلونوماید، به تنهایی یا در ترکیب با دیگر داروها، برای تعدیل دوره MS در افراد مبتلا به اشکال عود کننده، با یا بدون پیشرفت.
ویژگی‌های مطالعه
نویسندگان مرور، اثربخشی تری‌فلونوماید را عمدتا بر حسب تعداد شرکت‌کنندگان دچار حداقل یک عود، تعداد افراد با پیشرفت ناتوانی، نرخ سالانه عود (تعداد عود در سال به ازای هر شرکت‌کننده) و زمان سپری شده تا پیشرفت ناتوانی، در نظر گرفتند. آنها بی‌خطری را بر اساس تعداد شرکت‌کنندگان دچار عوارض جانبی، تعداد شرکت‌کنندگان دچار عوارض جانبی جدی، و تعداد شرکت‌کنندگانی که به دلیل عوارض جانبی در یک یا دو سال از مطالعه انصراف دادند یا خارج شدند، ارزیابی کردند. میان متون علمی مربوطه، پنج مطالعه معیارهای ورود را داشتند. آنها شامل 3231 شرکت‌کننده بوده و اثربخشی و بی‌خطری تری‌فلونوماید را به تنهایی یا همراه با داروی دیگری به نام اینترفرون‐بتا (interferon‐β; IFNβ)، در برابر دارونما (داروی ساختگی) یا IFNβ‐1a ارزیابی کردند. شواهد تا سپتامبر 2015 بروز است.
نتایج کلیدی
نویسندگان شواهدی را با کیفیت پائین یافتند که نشان می‌دهد هر دو دوز تری‌فلونوماید وقوع عود را پس از یک یا دو سال درمان کاهش می‌دهد، در حالی که شواهد با کیفیت بسیار پائین حاکی از آن است که این دارو از پیشرفت ناتوانی در یک یا دو سال پیشگیری می‌کند. تری‌فلونوماید با دوز بالا به جای دوز کم، اثربخشی مشابه با IFNβ‐1a از نظر کاهش عود در یک سال داشت، اما کیفیت شواهد بسیار پائین بود. تا آنجا که به بی‌خطری مربوط می‌شود، شایع‌ترین عوارض جانبی گزارش شده عبارت بودند از اسهال (مدفوع شل و مکرر)، تهوع (احساس بیماری)، نازک شدن مو، نوتروپنی (سطوح پائین گلبول‌های سفید خون به نام نوتروفیل‌ها، که با عفونت مبارزه می‌کنند) و لنفوپنی ( سطوح پائین گلبول‌های سفید خون به نام لنفوسیت‌ها که با عفونت مبارزه می‌کنند). به‌طور کلی، این عوارض جانبی خفیف تا متوسط بوده و معمولا منجر به توقف درمان نمی‌شوند، اما دوز بالاتر بیشتر مستعد ایجاد این عوارض جانبی است.
کیفیت شواهد
کیفیت پائین/بسیار پائین نتایج عمدتا به دلیل کورسازی (blinding) ناکافی ارزیابی عود (ارزیابان از درمان دریافت شده توسط فرد آگاه بودند)، نرخ بالای ترک مطالعه (افرادی که کارآزمایی را ترک کردند)، پیشرفت ناتوانی تایید شده در کمتر از شش ماه، کم بودن تعداد شرکت‌کنندگان، و طول درمان‌های مختلف طی مطالعات بود. طول دوره انجام مطالعات یک نقطه کلیدی برای یک بیماری مادام‌العمر با احتمال درمان‌های مزمن به عنوان MS است، هم‌چنین نیاز به انجام مطالعاتی را با دوره پایش طولانی‌تر (پیگیری) نشان می‌دهد. پنج مطالعه گنجانده شده در این مرور توسط شرکت‌های داروسازی حمایت مالی شدند و این به عنوان منبع بالقوه تضاد منافع و در نتیجه سوگیری (bias) شناخته می‌شود.
Authors' conclusions
Summary of findings
| Teriflunomide compared to placebo for multiple sclerosis | ||||||
| Patient or population: people with relapsing multiple sclerosis | ||||||
| Outcomes | Illustrative comparative risks* (95% CI) | Relative effect | No of Participants | Quality of the evidence | Comments | |
| Assumed risk | Corresponding risk | |||||
| Placebo | Teriflunomide | |||||
| Proportion of participants with at least 1 relapse at 1 year | 394 per 1000 | 237 per 1000 | RR 0.60 | 761 | ⊕⊕⊝⊝ | This outcome was considered low, because we considered there were very serious limitation in study design and execution. The bias that influenced the validity of the results for this outcome included: the high risks of bias due to unblinded assessments for relapse and conflicts of interest (sensitivity analysis according to a likely‐case scenario showed a robustness for the results of this outcome, we considered that the high attrition bias did not influence the robustness of the results on relapse). Therefore, we downgraded the quality of evidence for this outcome by 2 levels |
| The proportion of participants with at least 1 relapse at 2 years | 545 per 1000 | 436 per 1000 | RR 0.80 | 722 | ⊕⊕⊝⊝ | This outcome was considered low, because we considered there were very serious limitation in study design and execution. The bias that influenced the validity of the results for this outcome included: the high risks of bias due to unblinded assessments for relapse and conflicts of interest. (Sensitivity analysis according to a likely‐case scenario showed a robustness for the results of this outcome, we considered that the unclear attrition bias did not influence the robustness of the results on relapse.) Therefore, we downgraded the quality of evidence for this outcome by 2 levels |
| The proportion of participants with disability progression at 1 year | 142 per 1000 | 78 per 1000 | RR 0.55 | 761 | ⊕⊝⊝⊝ | This outcome was considered very low based on the following reasons:
|
| The proportion of participants with disability progression at 2 years | 273 per 1000 | 202 per 1000 | RR 0.74 | 722 | ⊕⊝⊝⊝ | This outcome was considered very low based on the following reasons:
|
| The proportion of participants with diarrhoea at 2 years | 89 per 1000 | 179 per 1000 | RR 2.01 | 718 | ⊕⊕⊕⊝ | The follow‐up periods were diverse in Confavreux 2014 and O'Connor 2011 (at least 48 weeks (Confavreux 2014) and 108 weeks (O'Connor 2011)). Treatment duration of participants in Confavreux 2014 was variable, ending 48 weeks after the last participant was included (a maximum treatment duration of 173 weeks). Actually, the data on adverse events in Confavreux 2014 were not at 2 years. There was a heterogeneity in follow‐up period between the studies. Therefore, we did not combine the data on adverse events in Confavreux 2014 and O'Connor 2011 |
| The proportion of participants with hair thinning at 2 years | 33 per 1000 | 131 per 1000 | RR 3.94 | 718 | ⊕⊕⊕⊝ | The follow‐up periods were diverse in Confavreux 2014 and O'Connor 2011 (at least 48 weeks (Confavreux 2014) and 108 weeks (O'Connor 2011)). Treatment duration of participants in Confavreux 2014 was variable, ending 48 weeks after the last participant was included (a maximum treatment duration of 173 weeks). Actually, the data on adverse events in Confavreux 2014 were not at 2 years. There was a heterogeneity in follow‐up period between the studies. Therefore, we did not combine the data on adverse events in Confavreux 2014 and O'Connor 2011 |
| The proportion of participants with elevated ALT levels at 2 years | 67 per 1000 | 143 per 1000 | RR 2.14 | 718 | ⊕⊕⊕⊝ | The follow‐up periods were diverse in Confavreux 2014 and O'Connor 2011 (at least 48 weeks (Confavreux 2014) and 108 weeks (O'Connor 2011)). Treatment duration of participants in Confavreux 2014 was variable, ending 48 weeks after the last participant was included (a maximum treatment duration of 173 weeks). Actually, the data on adverse events in Confavreux 2014 were not at 2 years. There was a heterogeneity in follow‐up period between the studies. Therefore, we did not combine the data on adverse events in Confavreux 2014 and O'Connor 2011 |
| *The basis for assumed risk is the placebo group risk. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). The assumed risk was defined as placebo group risk because only one study was evaluated. | ||||||
| GRADE Working Group grades of evidence | ||||||
| a High risks of bias existed in Confavreux 2014 due to unblinded assessments for relapse and conflicts of interest. b High risks of bias existed in O'Connor 2011 due to unblinded assessments for relapse and conflicts of interest. c High risks of bias existed in Confavreux 2014 due to effects of the high attrition bias on progression disability and conflicts of interest. d High risk of bias existed in O'Connor 2011 due to effects of the unclear attrition bias on progression disability and conflicts of interest. e Serious indirectness existed in Confavreux 2014 or in O'Connor 2011 because disability progression was confirmed at 3 months of follow‐up. f High risk of bias existed in O'Connor 2011 due to an unclear attrition bias and conflicts of interest. | ||||||
Background
Description of the condition
Multiple sclerosis (MS) is a chronic immune‐mediated disease of the central nervous system. It is pathologically characterized by inflammation, demyelination, and axonal and neuronal loss. Clinically it is characterized by recurrent relapses or progression, or both, typically striking adults during the primary productive time of their lives and ultimately leading to severe neurological disability.
In 1996, the clinical course of MS was characterized as relapsing‐remitting, primary progressive, secondary progressive or progressive relapsing. Initially, more than 80% of individuals with MS experience a relapsing‐remitting disease course (RRMS) characterized by clinical exacerbations of neurological symptoms followed by complete or incomplete remission (Lublin 1996). After 10 to 20 years, or median age of 39.1 years, about half of them gradually accumulate irreversible neurological deficits with or without clinical relapses (Confavreux 2006), which is known as secondary progressive MS (SPMS). Another 10% to 20% of individuals with MS are diagnosed with primary progressive MS (PPMS), clinically defined as a disease course without any clinical attacks or remission from onset (Lublin 1996). A significantly rarer form is progressive relapsing MS (PRMS), which initially presents as PPMS, however, during the course of the disease, these individuals develop true neurological exacerbations (Tullman 2004). In 2013, the clinical course of MS was re‐defined. In the new revisions, clinically isolated syndrome was added, and PRMS was eliminated, from the clinical course descriptions. All forms of MS should be further subcategorized as either active or non‐active. Active MS is defined as the occurrence of clinical relapse or the presence of new T2 or gadolinium‐enhancing lesions over a specified period of time, preferably at least one year. An additional subcategory for people with progressive MS differentiates between people who have shown signs of disability progression over a given time period and people who have remained stable (Lublin 2014a; Lublin 2014b).
MS causes a major socioeconomic burden, both for the individual and for society. Increased economic and quality of life (QoL) burden is associated with disease progression and relapses (Karampampa 2012; O'Connell 2014; Parisé 2013). From a person's perspective, an MS relapse is associated with a significant increase in economic costs as well as a decline in health‐related quality of life (HRQoL) and functional ability (Oleen‐Burkey 2012). Effective treatment that reduces relapse frequency and prevents progression could have an impact both on costs and HRQoL, and may help to reduce the social burden of MS (Karampampa 2012).
Description of the intervention
Teriflunomide, the active metabolite of leflunomide, is known to possess both anti‐proliferative and anti‐inflammatory actions. Data from human trials of leflunomide in rheumatoid arthritis showed that teriflunomide demonstrated linear pharmacokinetics over a dose range of 5 mg/day to 25 mg/day. The mean plasma half‐life is 15 days to 18 days and teriflunomide is extensively (greater than 99%) protein bound and exhibits linear protein binding at therapeutic concentrations. Clearance is via biliary and renal routes so administration of cholestyramine can be used to facilitate rapid elimination of teriflunomide from the circulation (Tallantyre 2008). Teriflunomide decreases disease severity and reduces inflammation, demyelination and axonal loss in a dose‐dependent manner in the Dark Agouti rat model of experimental autoimmune encephalomyelitis (EAE) (Merrill 2009). Teriflunomide (Aubagio®) was approved by the US Food and Drug Administration (FDA) in 2012 for people with relapsing forms of MS (7 mg or 14 mg orally once daily). In 2013, it was approved by the European Medicines Agency (EMA) for adults with RRMS (the recommended dose: 14 mg once a day).
How the intervention might work
Teriflunomide has an ability to non‐competitively and reversibly inhibit the mitochondrial enzyme dihydro‐orotate dehydrogenase (DHODH), a key cellular enzyme involved in the de novo synthesis of pyrimidine (Bruneau 1998; Greene 1995). By inhibiting DHODH and diminishing deoxyribonucleic acid (DNA) synthesis, teriflunomide has a cytostatic effect on proliferating B and T lymphocytes (Cherwinski 1995). Teriflunomide also inhibits protein tyrosinekinase activity (Xu 1996), resulting in the reduction of T‐cell proliferation, T‐cell production of interferon gamma (IFN‐γ) and interleukin 2 (IL2), as well as B‐cell immunoglobulin (Ig)G1 production and inhibition of nuclear factor ĸB (NFĸB) (Manna 1999; Siemasko 1998; Xu 1995). In addition, teriflunomide diminishes the ability of antigen‐presenting cells (APC) to activate T cells and for stimulated T cells to activate monocytes in vitro (Zeyda 2005), and inhibits interleukin 1 beta, matrix metalloproteinases (Deage 1998), and cyclo‐oxygenase‐2 activity (Hamilton 1999). In EAE, teriflunomide reduces activation of myelin basic protein (MBP)‐specific T cells then reduces the production of IFN‐γ and chemotaxis (Korn 2004).
Why it is important to do this review
This is an update of the Cochrane review "Teriflunomide for multiple sclerosis" (first published in The Cochrane Library 2012 Issue 12).
Objectives
To assess the absolute and comparative effectiveness and safety of teriflunomide as monotherapy or combination therapy versus placebo or other disease‐modifying drugs (DMDs) (interferon beta (IFNβ), glatiramer acetate, natalizumab, mitoxantrone, fingolimod, dimethyl fumarate, alemtuzumab) for modifying the disease course in people with MS.
Methods
Criteria for considering studies for this review
Types of studies
All randomized, controlled, parallel‐group clinical trials (RCTs) evaluating teriflunomide, as monotherapy or combination therapy, versus placebo or any approved DMDs for people with MS. We excluded trials with a length of follow‐up shorter than one year.
Types of participants
We included participants aged 18 years or older with definite diagnoses of MS according to Poser's (Poser 1983) or Mc Donald's (McDonald 2001; Polman 2005; Polman 2011) criteria, any clinical phenotypes categorized according to the classification of Lublin and Reingold (Lublin 1996), and an Expanded Disability Status Scale (EDSS) scores of 6.0 or lower.
Types of interventions
Experimental intervention
Treatment with teriflunomide orally, as monotherapy or combination therapy, without restrictions regarding dose, administration frequency and duration of treatment.
Control intervention
Placebo or an approved DMDs.
Types of outcome measures
Primary outcomes
Efficacy
-
The proportion of participants with at least one relapse at one year or two years. Confirmed relapse was defined as the occurrence of new symptoms or worsening of previously stable or improving symptoms and signs not associated with fever or infection that occurred at least 30 days after the onset of a preceding relapse and lasted more than 24 hours. The relapse should be verified by the examining neurologist within seven days after its occurrence and be accompanied by an increase of at least half a point in the EDSS score or at least one point in two functional systems (excluding change in sphincteric or cerebral functions).
-
The proportion of participants with disability progression as assessed by the EDSS (Kurtzke 1983) at one year or two years. Disability progression was defined as an increase in the EDSS score of at least 1.0 point in participants with a baseline score of 1.0 or higher or an increase of at least 1.5 points in participants with a baseline score of 0, with the increased score sustained for six months. We used the data where disability progression was confirmed in less than six months, however, we downgraded the study for indirectness of evidence when we performed the GRADE assessment.
Safety
-
The number of participants with adverse events (AEs), number of participants with serious adverse events (SAEs), and number of participants who withdrew or dropped out from the study because of AEs at one year or two years.
Secondary outcomes
-
The annualized relapse rate at one year or two years, defined as the mean number of confirmed relapses per participant adjusting for the duration of follow‐up to annualize it.
-
The number of gadolinium‐enhancing T1‐weighted lesions at one year or two years. Lesions that persisted for more than four weeks were counted more than once.
-
The time to disability progression at one year or two years.
-
Changes in T1 hypointensity or magnetization transfer ratio of lesion damage at one year or two years.
-
Mean change in HRQoL. The following scales were accepted: 36‐item Short Form (SF‐36) scores (Ware 1992), Multiple Sclerosis Quality of Life (MSQoL‐54) questionnaire scores (Vickrey 1995), Multiple Sclerosis Quality of Life Inventory (MSQLI) (Fischer 1999), or Functional Assessment of Multiple Sclerosis (FAMS) (Cella 1996) at one year or two years.
Search methods for identification of studies
We applied no language restrictions to the search.
Electronic searches
The Trials Search Co‐ordinator searched the Cochrane Multiple Sclerosis and Rare Diseases of the CNS Group Specialised Trials Register (30 September 2015), which, among other sources, contains trials from:
-
the Cochrane Central Register of Controlled Trials (CENTRAL) (2015 Issue 9);
-
MEDLINE (PubMed) (1966 to 30 September 2015);
-
EMBASE (EMBASE.com) (1974 to 30 September 2015);
-
Cumulative Index to Nursing and Allied Health Literature (CINAHL) (EBSCOhost) (1981 to 30 September 2015);
-
Latin American and Caribbean Health Science Information Database (LILACS) (Bireme) (1982 to 30 September 2015);
-
Clinical trial registries (clinicaltrials.gov);
-
World Health Organization (WHO) International Clinical Trials Registry Portal (apps.who.int/trialsearch/).
Information on the Trials Register and details of search strategies used to identify trials is in the 'Specialised Register' section within the Cochrane Multiple Sclerosis and Rare Diseases of the CNS Group's module.
Appendix 1 shows the keywords used to search for trials for this review.
Searching other resources
We checked the reference lists of published reviews and retrieved articles for additional trials. We searched reports (2004 to September 2015) from the MS Societies (National Multiple Sclerosis Society (United States, United Kingdom)) (www.nationalmssociety.org) and the Congress of the European Committee for Treatment and Research in Multiple Sclerosis (ECTRIMS (www.ectrims.eu) and Americas Committee for Treatment and Research in Multiple Sclerosis (ACTRIMS (www.actrims.org)). We communicated with investigators participating in trials of teriflunomide. We also contacted the Sanofi‐Aventis company in an effort to identify further studies (en.sanofi.com).
Data collection and analysis
Selection of studies
Two review authors (DH and YZ) independently screened titles and abstracts of the citations retrieved by the literature search for inclusion or exclusion. We obtained the available full texts of potentially relevant studies for further assessment. We independently evaluated the eligibility of these studies (on the basis of information available in the published data) and listed papers that did not meet the inclusion criteria in the Characteristics of excluded studies table with the reasons for exclusion. We resolved any disagreement regarding inclusion by discussion or by referral to a third review author (LC) if necessary.
Data extraction and management
Two review authors (DH and YZ) independently extracted information and data from the selected trials using standardized forms, including information about eligibility criteria, methods
(study design, total study duration, sequence generation, allocation sequence concealment, blinding and other concerns about bias), participants (total number, setting, diagnostic criteria, age, sex and country), interventions (total number of intervention groups and specific intervention) and outcomes (outcomes and time points, outcome definition and unit of measurement), results (number of participants allocated to each intervention group, sample size, missing participants and summary data for each intervention group) and funding source. Where the standard deviation was not reported, we calculated it from the standard error, confidence interval, t values or P values. We contacted the principal investigators of included studies to request additional data or confirmation of methodological aspects of the study. We discussed and resolved disagreements by consensus among the review authors.
Assessment of risk of bias in included studies
We assessed the risk of bias of the included studies using the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). Two review authors (DH and YZ) independently evaluated each study using the 'Risk of bias' tool under the domains of sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, selective outcome and other biases. We judged a study to have a high risk of attrition bias if it had a dropout rate higher than 20%, or the reasons for drop‐outs were not balanced across intervention groups. We judged a study to be at high risk of bias if at least one of the seven domains was rated at high risk of bias. Conversely, we judged a study to be at low risk of bias if all key domains were rated at low risk of bias, unless one or more of the domains was reported at unclear risk of bias, in which case we judged the study to be at unclear risk of bias.
Measures of treatment effect
We calculated the treatment effects of interventions based on the available data in the original studies using the Review Manager 5 analysis software (RevMan 2015). For dichotomous outcomes, such as the proportion of participants with at least one relapse, disability worsening and at least one AE, we used the risk ratio (RR) as the measure of treatment effect. We also calculated the risk difference (RD) (also called the absolute risk reduction) and the number needed to treat for an additional beneficial outcome (NNTB) or the number needed to treat for an additional harmful outcome (NNTH) (NNTB = 1/RD). We used the rate ratio as the measure of treatment effect for count data, such as the numbers of relapses and new gadolinium‐enhancing T1‐weighted lesions. Changes in T1 hypointensity and magnetization transfer ratio of lesion damage were continuous outcomes and we used the mean difference (MD) as the measure of treatment effect. We treated the data on QoL scales as continuous because they were longer ordinal rating scales and had a reasonably large number of categories. Therefore, we used the MD for trials that used the same rating scale. The time to disability progression was a time‐to‐event outcome, we summarized such data using methods of survival analysis and expressed the treatment effect as a hazard ratio (HR). We calculated 95% CIs for each treatment effect.
Unit of analysis issues
Most RCTs on teriflunomide for MS are multi‐arm studies with two experimental intervention groups (7 mg/day or 14 mg/day of teriflunomide) and a common control group, and involving repeated observations on participants. In future updates, where data are presented for each of the groups to which participants were randomized, we will create two pair‐wise comparisons of intervention groups to conduct independent meta‐analyses (high‐dose dimethyl fumarate group versus placebo group; low‐dose dimethyl fumarate group versus placebo group). Where outcomes are measured at multiple time points, we will define time frames to reflect short‐term (one year) and long‐term (two years) follow‐up.
Dealing with missing data
We did not conduct meta‐analyses because of the clinical and methodological diversity across the included studies. We included intention‐to‐treat data. We analysed the available data when the missing data can be reasonably assumed to be missing at random, but for data not missing at random, we performed sensitivity analyses according to a likely‐case scenario analysis, in which we assumed that both participants who dropped out both in the experimental group and in the control group had poor outcomes. We addressed the potential impact of missing data on the findings of the review in the Discussion section.
Assessment of heterogeneity
We assessed clinical heterogeneity by examining the characteristics of the studies and the similarity between the types of participants, interventions and outcomes. We also evaluated the variability in study design and risk of bias (methodological heterogeneity). We found obvious clinical and methodological heterogeneity across the included studies. If further data become available, we will evaluate statistical heterogeneity where clinical and methodological heterogeneity are not obvious across the included studies. When pooling trials in meta‐analyses, we will calculate the I2 statistic in order to identify heterogeneity across studies. An I2 value higher than 30% may indicate moderate heterogeneity (Higgins 2011).
Assessment of reporting biases
The trials included in this review did not permit an assessment of publication bias. If we include a sufficient number of RCTs in meta‐analysis (10 or more RCTs) in future updates, we will examine potential publication bias using a funnel plot. For continuous outcomes, we will use the standard error as the vertical axis and MDs as the horizontal axis in funnel plots. For dichotomous outcomes, we will plot RRs on a logarithmic scale as the horizontal axis and use the standard error as the vertical axis.
Data synthesis
We could not combine the outcome data because of the different study designs and interventions across studies; instead, we gave a descriptive summary of the results in the original studies. If we consider studies to be sufficiently clinically and methodologically similar in future updates, we will conduct formal meta‐analysis using Review Manager 5 software (RevMan 2015). We will conduct separate analyses in which higher‐dose teriflunomide (14 mg once daily) and lower‐dose teriflunomide (7 mg once daily) are compared to placebo. We will examine one‐year and two‐year outcomes separately. If we consider that all studies in a meta‐analysis are likely to be estimating the same underlying treatment effect, then we will use a fixed‐effect model, otherwise we will use a random‐effects model for meta‐analysis. For dichotomous outcomes, we will use the Mantel‐Haenszel method (Greenland 1985; Mantel 1959). For continuous outcomes, we will use the inverse‐variance method (DerSimonian 1986).
Subgroup analysis and investigation of heterogeneity
We could not carry out subgroup analysis because of the lack of data, but in future updates and if further data become available, we intend to undertake subgroup analyses according to:
-
different types of MS (e.g. people with RRMS or people with progressive MS);
-
baseline EDSS scores (e.g. 3.5 or lower, between 3.5 and 6);
-
different duration of MS (e.g. five years, more than five years);
-
risk of bias in included studies.
Sensitivity analysis
We undertook sensitivity analyses to assess the robustness of our review results. We conducted sensitivity analyses according to a likely‐case scenario in order to assess the effect of study withdrawal on the primary outcomes. Based on the intention‐to‐treat principle, we included all randomly assigned participants (including those who did not receive study treatment) into sensitivity analysis.
Results
Description of studies
See: Characteristics of included studies; Characteristics of excluded studies and Characteristics of ongoing studies.
Results of the search
In total, the search strategy retrieved 193 records after we removed duplicates. After screening of titles and abstracts, we selected seven studies reported in 42 articles provisionally and obtained the full papers for further assessment for eligibility. We excluded two studies (reported in 14 articles) due to a length of follow‐up shorter than one year or participants without a diagnosis of definite MS (Miller 2014; O'Connor 2006). Five studies met the inclusion criteria (Confavreux 2014; Freedman 2012; NCT01252355; O'Connor 2011; Vermersch 2014) (reported in 28 articles, the results of NCT01252355 were published on clinicaltrials.gov). We included the trials that we classified in ongoing studies in previous versions of this review (NCT00751881 and NCT01252355) in the current review (Confavreux 2014; NCT01252355). See Figure 1.
Study flow diagram.
Included studies
The review included five studies involving 3231 people (Confavreux 2014; Freedman 2012; NCT01252355; O'Connor 2011; Vermersch 2014). Among them, two studies evaluated the efficacy and safety of teriflunomide 7 mg/day or 14 mg/day versus placebo for 2257 adults with relapsing forms of MS (Confavreux 2014; O'Connor 2011). Two studies primarily evaluated the safety and tolerability of teriflunomide 7 mg/day or 14 mg/day with add‐on IFNβ versus placebo in 650 people with relapsing MS (Freedman 2012; NCT01252355). One study evaluated the efficacy, safety and tolerability of teriflunomide 7 mg/day or 14 mg/day in comparison to IFNβ‐1a in 324 people with relapsing MS (Vermersch 2014).
Characteristics of the study design
Confavreux 2014 and O'Connor 2011 were randomized, double‐blind, placebo‐controlled, parallel‐group studies over at least 48 weeks (a maximum of 173 weeks) (Confavreux 2014) and 108 weeks (O'Connor 2011). Freedman 2012 was a randomized, placebo‐controlled, 24‐week double‐blind study followed by a 24‐week blinded extension. Participants completing 24 weeks of treatment who continued to meet the eligibility criteria could select to enter a 24‐week blinded extension during which participants continued to receive their originally assigned treatment regimen. NCT01252355 was a randomized, double‐blind, placebo‐controlled, parallel‐group study over 24 weeks (a maximum of 108 weeks). Vermersch 2014 was an approved DMD‐controlled, parallel‐group, rater‐blinded study over at least 48 weeks (a maximum of 115 weeks).
Characteristics of the participants
All participants had a diagnosis of definite MS according to McDonald's diagnostic criteria (McDonald 2001; Polman 2005), an age ranging from 18 to 55 years and a relapsing clinical course with or without progression (RRMS, SPMS or PRMS). All participants had an entry score of 5.5 or lower on the EDSS and no relapse for at least 30 days before randomization. The participants in Confavreux 2014 and O'Connor 2011 had at least one relapse in the previous year or at least two clinical relapses in the previous two years. The participants in NCT01252355 had disease activity in the one year prior to randomization and after first three months of IFNβ treatment. Baseline demographic and disease characteristics were well balanced among the groups in most studies except for Vermersch 2014, in which DMD use in the past two years in the teriflunomide 14 mg/day group was lower than that in the IFNβ‐1a group.
Characteristics of the interventions
Participants in Confavreux 2014 and O'Connor 2011 received oral teriflunomide 7 mg once daily or oral teriflunomide 14 mg once daily or a matching placebo for at least 48 weeks (core treatment period: 48 weeks to 152 weeks, a maximum of 173 weeks) (Confavreux 2014) and 108 weeks (O'Connor 2011). Participants in Freedman 2012 and NCT01252355 received oral administration of 7 mg/day of teriflunomide added to IFNβ, 14 mg/day of teriflunomide added to IFNβ or matching placebo added to IFNβ for 48 weeks (Freedman 2012) and at least 24 weeks (a maximum of 108 weeks) (NCT01252355). Participants in Vermersch 2014 received oral teriflunomide 7 mg once daily or oral teriflunomide 14 mg once daily or IFNβ‐1a 44 μg by subcutaneous injection three times per week for at least 48 weeks (a maximum of 115 weeks).
Characteristics of the outcome measures
All studies reported the proportion of participants with at least one relapse. Two studies reported sustained disability progression, which was defined as an increase from baseline of at least 1.0 point in the EDSS score (or at least 0.5 points for participants with a baseline EDSS score greater than 5.5) that persisted for at least 12 weeks (Confavreux 2014; O'Connor 2011). All studies reported the number of participants with AEs, number of participants with SAEs and number of participants who withdrew or dropped out from the study because of AEs.
All studies reported the annualized relapse rate. Three trials reported the number of gadolinium‐enhancing T1‐weighted lesions (Freedman 2012; NCT01252355; O'Connor 2011). Three trials reported the time to disability progression (Confavreux 2014; NCT01252355; O'Connor 2011). One trial reported changes in T1 hypointensity of lesion damage (O'Connor 2011). Two trials reported mean change in QoL measured by SF‐36 scores (Confavreux 2014; NCT01252355). None of the studies reported magnetization transfer ratio of lesion damage. One trial did not provide data at one year (O'Connor 2011).
Excluded studies
We excluded two studies (reported in 14 articles) from this review; the reasons for their exclusion are listed in the Characteristics of excluded studies table.
Risk of bias in included studies
Further details of this assessment are available in the Characteristics of included studies table and are also presented in the 'Risk of bias' graph (Figure 2) and 'Risk of bias' summary (Figure 3).
Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.
Risk of bias summary: review authors' judgements about each risk of bias item for each included study.
Allocation
In Confavreux 2014 and O'Connor 2011, sequence generation and allocation concealment were adequate. Allocation sequence was generated by randomization number list (Confavreux 2014) or a permuted‐block randomization schedule with stratification (O'Connor 2011). Randomization was done centrally, via an interactive voice recognition system (IVRS) in both studies. In NCT01252355, sequence generation was probably made by software. Assignment to groups was done centrally using an IVRS. In Freedman 2012 and Vermersch 2014, sequence generation was probably made by software, and central randomization was probably used.
Blinding
In Confavreux 2014 and O'Connor 2011, the treating neurologist who recorded adverse events was responsible for assessment of relapses, blinding of relapse assessment was probably not adequate. The risk of detection bias was high. Participants included in Freedman 2012 and NCT01252355 received diverse regimens of IFNβ, they were not truly double‐blind, double‐dummy studies. The control (IFNβ‐1a) group was open‐label in Vermersch 2014, it was not a truly double‐blind study. In addition, the treating neurologist who reported or managed adverse events was responsible for assessment of relapses in Freedman 2012, NCT01252355, and Vermersch 2014, blinding of relapse assessment was probably not adequate. The risks of performance bias and detection bias were high.
Incomplete outcome data
Three studies had a high risk of attrition bias due to a high dropout rate of 29.8% (Confavreux 2014), 36.4% (Freedman 2012), and 100% (NCT01252355). There was an overall dropout rate of 20.1% in O'Connor 2011, but there was no sufficient information to understand the reasons for study discontinuation and their balance among the groups. The risk of attrition bias was unclear. One study did not report the number and reasons of drop‐outs and the incomplete outcome data were unclear (Vermersch 2014).
Selective reporting
All studies reported all listed outcomes adequately.
Other potential sources of bias
All studies were sponsored by Sanofi‐Aventis. In Confavreux 2014 and O'Connor 2011, the sponsor analysed the data and some co‐authors were affiliated to Sanofi‐Aventis. In Vermersch 2014, Sanofi‐Aventis funded editorial support. Conflicts of interest were obvious, and there was a high risk of bias in all studies.
Effects of interventions
See: Summary of findings for the main comparison Teriflunomide compared to placebo for multiple sclerosis
We did not conduct meta‐analyses because of the high risk of bias and clinical diversities of the included studies. The study designs in Confavreux 2014 and O'Connor 2011 were similar, however, the follow‐up periods were diverse (at least 48 weeks (Confavreux 2014) and 108 weeks (O'Connor 2011)). Treatment duration in Confavreux 2014 was variable, ending 48 weeks after the last participant was included (a maximum treatment duration of 173 weeks). Furthermore, the data at one year in O'Connor 2011 were unavailable. The study designs in Freedman 2012 and NCT01252355 were also similar, but the follow‐up periods were diverse (48 weeks (Freedman 2012) and at least 24 weeks (NCT01252355)). Treatment duration in NCT01252355 was variable (24 weeks to 108 weeks). Consequently, we could only calculate the treatment effects of interventions based on the available data in the original studies.
Primary outcomes
Efficacy: proportion of participants with at least one relapse at one year or two years
All studies reported proportion of participants with at least one relapse at one year or two years.
Confavreux 2014 reported the proportion of participants with at least one relapse at one year of follow‐up were 28.10% with low‐dose teriflunomide, 23.70% with high‐dose teriflunomide and 39.40% with placebo, and the RD was 11.30% with low‐dose teriflunomide and 15.70% with high‐dose teriflunomide. Compared to placebo, the results showed low dose of teriflunomide as monotherapy reduced the number of participants with at least one relapse at one year of follow‐up (RR 0.72, 95% CI 0.59 to 0.87, P value = 0.001; 797 participants) and the NNTB was 9, which means that they needed to treat nine participants with low‐dose teriflunomide to prevent one participant relapsing during the one years of follow‐up. Similarly, high dose of teriflunomide as monotherapy also reduced the number of participants with at least one relapse at one year of follow‐up (RR 0.60, 95% CI 0.48 to 0.75, P value < 0.00001; 761 participants) and the NNTB was 6, which means that they needed to treat six participants with high‐dose teriflunomide to prevent one participant relapsing during the one year of follow‐up. Assuming participants who withdrew from study both in experimental groups and control group had a relapse, the likely‐case scenario analyses showed both doses of teriflunomide reduced the number of participants with at least one relapse at one year of follow‐up (low dose: RR 0.83, 95% CI 0.75 to 0.93, P value = 0.0008, 797 participants; high dose: RR 0.79, 95% CI 0.70 to 0.88, P value < 0.0001, 761 participants).
Similarly, at two years of follow‐up in O'Connor 2011, compared to placebo, both doses of teriflunomide as monotherapy reduced the number of participants with at least one relapse at two years of follow‐up (low dose: RR 0.85, 95% CI 0.74 to 0.98, P value = 0.03; 729 participants; high dose: RR 0.80, 95% CI 0.69 to 0.93, P value = 0.004; 722 participants). The proportion of participants with at least one relapse at two years of follow‐up were 46.30% with low‐dose teriflunomide, 43.50% with high‐dose teriflunomide and 54.40% with placebo, and the RD was 8.10% with low‐dose teriflunomide and 10.90% with high‐dose teriflunomide, corresponding to an NNTB of 12 with low‐dose teriflunomide and 9 with high‐dose teriflunomide, which means that they needed to treat 12 participants with low‐dose teriflunomide, and nine participants with high‐dose teriflunomide to prevent one participant relapsing during the two years of follow‐up. When taking the effect of drop‐outs into consideration, the likely‐case scenario analyses still showed a benefit in reducing the number of participants with at least one relapse for both doses of teriflunomide (low dose: RR 0.88, 95% CI 0.80 to 0.97, P value = 0.008; 729 participants; high dose: RR 0.87, 95% CI 0.79 to 0.96, P value = 0.005; 722 participants).
Freedman 2012 showed neither doses of teriflunomide added to IFNβ were superior to placebo added to IFNβ concerning the proportion of participants with at least one relapse at one year of follow‐up (low dose: RR 1.08, 95% CI 0.45 to 2.59, P value = 0.86; 79 participants; high dose: RR 0.79, 95% CI 0.30 to 2.07, P value = 0.63; 80 participants). However, NCT01252355 showed opposite results, both doses of teriflunomide added to IFNβ were superior to placebo added to IFNβ concerning the proportion of participants with at least one relapse at one year of follow‐up (low dose: RR 0.60, 95% CI 0.42 to 0.87, P value = 0.007; 353 participants; high dose: RR 0.58, 95% CI 0.40 to 0.84, P value = 0.004; 354 participants). When administrated as monotherapy for 48 weeks to 115 weeks in Vermersch 2014, low dose of teriflunomide was inferior to IFNβ‐1a on the proportion of participants with at least one relapse (RR 2.74, 95% CI 1.66 to 4.53, P value < 0.0001; 213 participants), but there was no difference in reducing the number of participants with at least one relapse for high dose of teriflunomide (RR 1.52, 95% CI 0.87 to 2.67, P value = 0.14; 215 participants).
Efficacy: proportion of participants with disability progression
Confavreux 2014 reported the proportions of participants with progression of disability at one year of follow‐up were 12.10% with low‐dose teriflunomide, 7.80% with high‐dose teriflunomide and 14.20% with placebo, and the RD was 2.10% with low‐dose teriflunomide and 6.40% with high‐dose teriflunomide. Compared to placebo, the results showed high dose of teriflunomide as monotherapy reduced the number of participants with disability progression at one year of follow‐up (RR 0.55, 95% CI 0.36 to 0.84, P value = 0.006; 761 participants), and the NNTB was 16, which means that they needed to treat 16 participants with high‐dose teriflunomide to prevent one participant having disability progression during the one year of follow‐up. However, there was no difference for low dose of teriflunomide in disability progression at one year of follow‐up (RR 0.85, 95% CI 0.59 to 1.22, P value = 0.37; 797 participants). When taking the effect of drop‐outs into consideration, the likely‐case scenario analysis showed neither dose of teriflunomide reduced the number of participants with disability progression at one year of follow‐up (low dose: RR 0.94, 95% CI 0.80 to 1.11, P value = 0.47; 797 participants; high dose: RR 0.88, 95% CI 0.74 to 1.04, P value = 0.14; 761 participants).
O'Connor 2011 reported the risk of disability progression at two years of follow‐up was 21.70% with low‐dose teriflunomide and 20.20% with high‐dose teriflunomide lower than that in participants receiving placebo (27.3%). The RD was 5.60% with low‐dose teriflunomide and 7.10% with high‐dose teriflunomide. Compared to placebo, the results showed high dose of teriflunomide as monotherapy reduced the proportion of participants with disability progression at two years of follow‐up (RR 0.74, 95% CI 0.56 to 0.96, P value = 0.02; 722 participants), and the NNTB was 14, which means that they needed to treat 14 participants with high‐dose teriflunomide to prevent one participant against disability progression during the two years of follow‐up. However, there was no difference for low dose of teriflunomide (RR 0.79, 95% CI 0.61 to 1.02, P value = 0.08; 729 participants). However, the likely‐case scenario analysis showed neither dose of teriflunomide reduced the number of participants with disability progression at two years of follow‐up (low dose: RR 0.89, 95% CI 0.75 to 1.06, P value = 0.20; 729 participants; high dose: RR 0.92, 95% CI 0.77 to 1.09, P value = 0.32; 722 participants).
Safety
Confavreux 2014 reported the safety of teriflunomide as monotherapy after the core treatment period of 48 weeks to 152 weeks. Compared to placebo, there was no difference for both doses of teriflunomide in the incidence of AEs (low dose: RR 1.01, 95% CI 0.95 to 1.08, P value = 0.71; 794 participants; high dose: RR 1.04, 95% CI 0.98 to 1.10, P value = 0.23; 756 participants) or SAEs (low dose: RR 1.04, 95% CI 0.72 to 1.51, P value = 0.83; high dose: RR 0.97, 95% CI 0.66 to 1.43, P value = 0.88). However, the incidence of AEs leading to discontinuation of the study medication in both teriflunomide groups was higher than that in the placebo group (low dose: RR 2.08, 95% CI 1.31 to 3.30, P value = 0.002; NNTH 15; high dose: RR 2.51, 95% CI 1.59 to 3.95, P value < 0.0001; NNTH 11). The most common AEs with an increased incidence in both teriflunomide groups included hair thinning (low dose: RR 2.33, 95% CI 1.35 to 4.01, P value = 0.002; NNTH 17; high dose: RR 3.05, 95% CI 1.79 to 5.19, P value < 0.0001; NNTH 11), neutropenia (low dose: RR 2.48, 95% CI 1.26 to 4.90, P value = 0.009; NNTH 24; high dose: RR 3.30, 95% CI 1.70 to 6.40, P value = 0.0004; NNTH 15), neutrophil counts less than 1.5 x 109/L (low dose: RR 1.85, 95% CI 1.18 to 2.90, P value = 0.008; NNTH 18; high dose: RR 2.47, 95% CI 1.60 to 3.82, P value < 0.001; NNTH 10), lymphocyte counts less than 0.8 x 109/L (low dose: RR 1.82, 95% CI 1.18 to 2.80, P value = 0.007; NNTH 17; high dose: RR 1.78, 95% CI 1.14 to 2.77, P value = 0.01; NNTH 18), and elevated alanine aminotransferase (ALT) levels greater than one time the upper limit of the normal range (low dose: RR 1.30, 95% CI 1.11 to 1.53, P value = 0.001; NNTH 9; high dose: RR 1.44, 95% CI 1.23 to 1.68, P value < 0.00001; NNTH 6).
There was a similar incidence of elevated ALT levels three times or greater the upper limit of the normal range and neutrophil counts less than 0.5 x 109/L between the placebo group and the teriflunomide groups. Elevated ALT and lymphocyte counts less than 0.5 x 109/L occurred at higher frequency with high‐dose teriflunomide compared to placebo (elevated ALT: RR 1.69, 95% CI 1.11 to 2.56, P value = 0.01; NNTH 18; elevated lymphocyte count: RR 11.42, 95% CI 1.48 to 87.98, P value = 0.02; NNTH 59), but there was no difference for low‐dose teriflunomide. In addition, diarrhoea was more common with low‐dose teriflunomide rather than high‐dose teriflunomide (RR 1.65, 95% CI 1.06 to 2.57, P value = 0.03). The proportion of other AEs most commonly reported in teriflunomide groups, such as headache, fatigue, nausea, nasopharyngitis, upper respiratory tract infection, back pain and urinary tract infection, were not higher than in the placebo group. The AEs leading to treatment discontinuation mainly included elevated ALT levels (3% with low‐dose teriflunomide and 2% with high‐dose teriflunomide), neutropenia (1% with low‐dose teriflunomide and 2% with high‐dose teriflunomide), hair thinning (2% with high‐dose teriflunomide) and diarrhoea (1% in both teriflunomide groups). There were 18 pregnancies in 14 female participants and four female partners of male participants. Of the 14 female participants, 10 elected to have induced abortions and four pregnancies resulted in healthy babies (one in the placebo group, two in the low‐dose teriflunomide group and one in the high‐dose teriflunomide group). Of the four pregnancies in partners of male participants, one woman elected to have an induced abortion and three pregnancies resulted in healthy babies (all in the low‐dose teriflunomide group).
O'Connor 2011 reported the safety of teriflunomide as monotherapy at two years of follow‐up. Compared to placebo, there was no difference for both doses of teriflunomide in the incidence of AEs (low dose: RR 1.02, 95% CI 0.97 to 1.07, P value = 0.49; 728 participants; high dose: RR 1.04, 95% CI 0.99 to 1.09, P value = 0.16; 718 participants), SAEs (low dose: RR 1.11, 95% CI 0.76 to 1.60, P value = 0.59; high dose: RR 1.25, 95% CI 0.87 to 1.79, P value = 0.23) and AEs leading to discontinuation of the study medication (low dose: RR 1.21, 95% CI 0.76 to 1.94, P value = 0.41; high dose: RR 1.35, 95% CI 0.86 to 2.14, P value = 0.20). The most common adverse events with an increased incidence in both teriflunomide groups included diarrhoea (low dose: RR 1.65, 95% CI 1.09 to 2.49, P value = 0.02; NNTH 17; high dose: RR 2.01, 95% CI 1.35 to 3.00, P value = 0.0006; NNTH 11), hair thinning or decreased hair density (low dose: RR 3.10, 95% CI 1.65 to 5.83, P value = 0.0005; NNTH 14; high dose: RR 3.94, 95% CI 2.13 to 7.30, P value < 0.0001; NNTH 10), elevated ALT levels (low dose: RR 1.79, 95% CI 1.11 to 2.89, P value = 0.02; NNTH 19; high dose: RR 2.14, 95% CI 1.35 to 3.39, P value = 0.001; NNTH 13). The incidence of nausea in high‐dose teriflunomide group rather than in low‐dose teriflunomide group was higher than that in placebo group (low dose: RR 1.24, 95% CI 0.76 to 2.03, P value = 0.39; high dose: RR 1.90, 95% CI 1.21 to 2.98, P value = 0.006; NNTH 15). The incidence of elevated ALT levels one times or greater the upper limit of the normal range in both doses was higher than that in placebo group (low dose: RR 2.61, 95% CI 1.23 to 5.53, P value = 0.01; NNTH 6; high dose: RR 3.24, 95% CI 1.56 to 6.75, P value = 0.002; NNTH 5), but there was no difference for both doses of teriflunomide in the incidence of elevated ALT levels three times or greater the upper limit of the normal range. These events rarely led to discontinuation of the study medication: diarrhoea (0.3% in both teriflunomide groups), nausea (0.3% with low‐dose), hair thinning or decreased hair density (0.5% with low‐dose and 1.4% with high‐dose). The proportion of other AEs (10% or greater) most commonly reported in any teriflunomide group, such as nasopharyngitis, headache, fatigue, influenza, back pain and urinary tract infection, occurred with a similar frequency in the placebo group. Mean reductions in neutrophil and lymphocyte counts from baseline values were small in magnitude (1.0 x 109/L or less for neutrophil counts and 0.3 x 109/L or less for lymphocyte counts) but were slightly more marked with high‐dose teriflunomide than with low‐dose teriflunomide or placebo. Moderate neutropenia (defined as a neutrophil count of less than 0.9 x 109/L) developed in three participants receiving teriflunomide. Eleven pregnancies occurred, leading to four spontaneous abortions (one in the placebo group and three in the high‐dose teriflunomide group), six induced abortions (five in the low‐dose teriflunomide group and one in the high‐dose teriflunomide group). One participant in the high‐dose teriflunomide group (treated for 31 days of the pregnancy) delivered a healthy baby with no reported health concerns after two years.
Vermersch 2014 reported the safety of teriflunomide as monotherapy after the core treatment period of 48 weeks to 115 weeks. Compared to IFNβ‐1a, there was no difference for both doses of teriflunomide in the incidence of AEs (low dose: RR 0.97, 95% CI 0.92 to 1.04, P value = 0.43; 211 participants; high dose: RR 0.97, 95% CI 0.90 to 1.03, P value = 0.29; 211 participants) or SAEs (low dose: RR 1.57, 95% CI 0.64 to 3.84, P value = 0.32; high dose: RR 0.79, 95% CI 0.27 to 2.26, P value = 0.66). However, the incidence of AEs leading to discontinuation in the IFNβ group was higher than those in the teriflunomide groups (low dose: RR 0.38, 95% CI 0.18 to 0.78, P value = 0.008; high dose: RR 0.50, 95% CI 0.26 to 0.96, P value = 0.04). The most commonly reported AEs (10% or greater) in either teriflunomide group were nasopharyngitis, headache, paraesthesia, diarrhoea, hair thinning, back pain and elevated ALT levels. Among these AEs, the incidence of diarrhoea in both teriflunomide groups was higher than that in the IFNβ‐1a group (low dose: RR 2.87, 95% CI 1.36 to 6.07, P value = 0.006; high dose: RR 2.64, 95% CI 1.24 to 5.63, P value = 0.01). Compared to IFNβ‐1a, hair thinning was more common with high‐dose teriflunomide rather than low‐dose teriflunomide (RR 20.20, 95% CI 2.77 to 147.14, P value = 0.003). However, elevated ALT levels occurred with a lower frequency in the teriflunomide groups (low dose: RR 0.36, 95% CI 0.19 to 0.65, P value = 0.0009; high dose: RR 0.33, 95% CI 0.17 to 0.61, P value = 0.0005). In addition, influenza‐like illness was reported more frequently with IFNβ‐1a than with teriflunomide (low dose: RR 0.07, 95% CI 0.03 to 0.18, P value < 0.00001; high dose: RR 0.05, 95% CI 0.02 to 0.16, P value < 0.00001). There was a similar incidence of other AEs between the IFNβ‐1a group and teriflunomide groups.
Freedman 2012 reported the safety of teriflunomide added to IFNβ at one year of follow‐up. Compared to placebo added to IFNβ, there was no difference for either dose of teriflunomide in the incidence of AEs (low dose: RR 1.11, 95% CI 0.96 to 1.29, P value = 0.17; 78 participants; high dose: RR 1.02, 95% CI 0.85 to 1.21, P value = 0.85; 79 participants), SAEs (low dose: RR 2.22, 95% CI 0.43 to 11.40, P value = 0.34; high dose: RR 0.54, 95% CI 0.05 to 5.71, P value = 0.61) and AEs leading to discontinuation (low dose: RR 1.66, 95% CI 0.29 to 9.40, P value = 0.57; high dose: RR 1.62, 95% CI 0.29 to 9.16, P value = 0.59). The most commonly reported AEs (10% or greater) in either teriflunomide group were elevated ALT levels, headache, decreased lymphocyte counts, nasopharyngitis, nausea, fatigue, decreased neutrophil counts, hypertension, back pain, vomiting, diarrhoea and urinary tract infection. However, these AEs occurred with a similar frequency in the IFNβ group.
NCT01252355 reported the safety of teriflunomide added to IFNβ after the treatment period of 28 weeks to 108 weeks. Compared to placebo added to IFNβ, the incidences of AEs in both teriflunomide groups were higher than those in the placebo group (low dose: RR 1.15, 95% CI 1.01 to 1.31, P value = 0.03; 354 participants; high dose: RR 1.16, 95% CI 1.02 to 1.31, P value = 0.02; 353 participants). However, there was no difference for either dose of teriflunomide in the incidence of SAEs (low dose: RR 1.59, 95% CI 0.68 to 3.74, P value = 0.29; high dose: RR 1.72, 95% CI 0.74 to 4.00, P value = 0.21). The incidence of AEs leading to discontinuation with high‐dose teriflunomide rather than with low‐dose teriflunomide was higher than that in the placebo group (low dose: RR 1.74, 95% CI 0.79 to 3.83, P value = 0.17; high dose: RR 2.40, 95% CI 1.14 to 5.07, P value = 0.02). There was no difference for either dose of teriflunomide in the incidence of elevated ALT levels three times or greater the upper limit of the normal range (low dose: RR 1.46, 95% CI 0.53 to 4.01, P value = 0.47; high dose: RR 1.47, 95% CI 0.53 to 4.03, P value = 0.46).
Secondary outcomes
Annualized relapse rate
All studies reported the annualized relapse rate. Confavreux 2014 reported annualized relapse rate after the core treatment period of 48 weeks to 152 weeks (low dose: annualized relapse rate 0.39, 95% CI 0.33 to 0.46; 407 participants; high dose: annualized relapse rate 0.32, 95% CI 0.27 to 0.38; 370 participants; placebo: annualized relapse rate 0.50, 95% CI 0.43 to 0.58; 388 participants), but we could not calculate the total number of relapses and standard errors due to the variable duration of follow‐up, consequently we could not calculate the rate ratio. However, the authors reported the RRs on annualized relapse rate, showing both doses of teriflunomide as monotherapy reduced annualized relapse rate during the follow‐up period of 48 weeks to 132 weeks (low dose: RR 0.78, 95% CI 0.63 to 0.96, P value = 0.0183; 797 participants; high dose: RR 0.64, 95% CI 0.51 to 0.79, P value = 0.0001; 761 participants).
There were similar results at two years of follow‐up in O'Connor 2011 (low dose: rate ratio 0.69, 95% CI 0.59 to 0.81, P value < 0.00001; 729 participants; high dose: rate ratio 0.69, 95% CI 0.59 to 0.80, P value < 0.00001; 722 participants).
Freedman 2012 showed neither doses of teriflunomide added to IFNβ were superior to placebo added to IFNβ concerning annualized relapse rate at one year of follow‐up (low dose: rate ratio 0.42, 95% CI 0.15 to 1.16, P value = 0.10; 79 participants; high dose: rate ratio 0.67, 95% CI 0.29 to 1.54, P value = 0.35; 80 participants).
NCT01252355 reported only the data of annualized relapse rate after the treatment duration of 24 weeks to 108 weeks (low dose: annualized relapse rate 0.242, 95% CI 0.152 to 0.386; 178 participants; high dose: annualized relapse rate 0.238, 95% CI 0.162 to 0.351; 179 participants; placebo: annualized relapse rate 0.298, 95% CI 0.206 to 0.432; 175 participants). Therefore, we could not calculate the total number of relapses and the standard error due to the variable duration of follow‐up, consequently we could not calculate the rate ratio.
Vermersch 2014 reported the data of annualized relapse rate after the treatment period of 48 weeks to 115 weeks (low dose: annualized relapse rate 0.41, 95% CI 0.27 to 0.64; 109 participants; high dose: annualized relapse rate 0.26, 95% CI 0.15 to 0.44; 111 participants; IFNβ‐1a: annualized relapse rate 0.22, 95% CI 0.11 to 0.42; 104 participants). However, we could not calculate the total number of relapses and the standard error due to the variable duration of follow‐up, consequently we could not calculate the rate ratio. However, the authors reported the RR on annualized relapse rate, showing that low‐dose teriflunomide was inferior to IFNβ‐1a on annualized relapse rate (RR 1.90, 95% CI 1.05 to 3.43, P value = 0.03; 213 participants), but there was no difference in reducing annualized relapse rate for high‐dose teriflunomide (RR 1.20, 95% CI 0.62 to 2.30, P value = 0.59; 215 participants).
Number of gadolinium‐enhancing T1‐weighted lesions
Three studies reported the number of gadolinium‐enhancing T1‐weighted lesions (Freedman 2012; NCT01252355; O'Connor 2011).
Compared to placebo, the results of O'Connor 2011 showed both doses of teriflunomide as monotherapy reduced the number of gadolinium‐enhancing T1‐weighted lesions at two years of follow‐up (low dose: rate ratio 0.43, 95% CI 0.37 to 0.51, P value < 0.00001; 729 participants; high dose: rate ratio 0.19, 95% CI 0.15 to 0.24, P value < 0.00001; 722 participants).
Freedman 2012 showed neither dose of teriflunomide added to IF‐β was superior to placebo added to IFNβ concerning the number of gadolinium‐enhancing T1‐weighted lesions at one year of follow‐up (low dose: rate ratio 0.67, 95% CI 0.29 to 1.55, P value = 0.35; 79 participants; high dose: rate ratio 0.42, 95% CI 0.16 to 1.09, P value = 0.08; 80 participants).
NCT01252355 reported only the data of the number of gadolinium‐enhancing T1‐weighted lesions after the treatment period of 24 weeks to 108 weeks (low dose: 0.257, 95% CI 0.127 to 0.523; 142 participants; high dose: 0.158, 95% CI 0.070 to 0.360; 151 participants; placebo: 0.542, 95% CI 0.344 to 0.855; 151 participants). We could not calculate the annualized relapse rate due to the variable duration of follow‐up, consequently we could not calculate the rate ratio.
Time to disability progression
Three studies reported the time to disability progression (Confavreux 2014; NCT01252355; O'Connor 2011).
Compared to placebo, the results of Confavreux 2014 showed high‐dose teriflunomide as monotherapy delayed the progression of disability after the core treatment period of 48 weeks to 152 weeks (HR 0.68, 95% CI 0.47 to 1.00, P value = 0.04; 758 participants), but there was no difference in delaying the progression of disability for low‐dose teriflunomide (HR 0.95, 95% CI 0.68 to 1.35, P value = 0.76; 795 participants).
The results of O'Connor 2011 showed high‐dose teriflunomide as monotherapy delayed the progression of disability at two years of follow‐up (HR 0.70, 95% CI 0.51 to 0.96, P value = 0.03; 721 participants), but there was no difference in delaying the progression of disability for low‐dose teriflunomide (HR 0.76, 95% CI 0.56 to 1.05, P value = 0.08; 728 participants).
Data of the time to disability progression in NCT01252355 were insufficient because of early study termination and were not reported in the original publication.
Changes in T1 hypointensity or magnetization transfer ratio of lesion damage
One study reported changes in T1 hypointensity lesion damage (O'Connor 2011). The results showed high‐dose teriflunomide as monotherapy, compared to placebo, reduced the volume of hypointense lesions on T1‐weighted images at two years (MD ‐0.20, 95% CI ‐0.35 to ‐0.05, P value = 0.009, 728 participants). However, there was no difference for low‐dose teriflunomide (MD ‐0.03, 95% CI ‐0.19 to 0.13, P value = 0.71; 721 participants).
None of the studies reported magnetization transfer ratio of lesion damage.
Change in health‐related quality of life
Two studies reported change in QoL measured by SF‐36 scores (Confavreux 2014; NCT01252355).
Confavreux 2014 found that compared to placebo, there was no difference for teriflunomide as monotherapy in change of SF‐36 physical health summary score and in SF‐36 mental health summary score at 48 weeks (low dose: physical health: MD 0.68, 95% CI ‐0.44 to 1.80, P value = 0.24 and mental health: MD 0.88, 95% CI ‐0.72 to 2.48, P value = 0.28; 797 participants; high dose: physical health: MD 0.97, 95% CI ‐0.18 to 2.12, P value = 0.10 and mental health: MD 1.48, 95% CI ‐0.18 to 3.14, P value = 0.08; 761 participants).
Data of change in QoL measured by SF‐36 scores in NCT01252355 were insufficient because of early study termination and were not reported in the original publication.
Discussion
Summary of main results
This systematic review included five RCTs involving 3231 adults with relapsing MS. All participants had a score of less than 5.5 on the EDSS and a relapsing clinical course with or without progression (RRMS, SPMS or PRMS). Two studies were large‐scale RCTs in which participants had disease activity with at least one relapse in the previous year or at least two clinical relapses in the previous two years. These two studies primarily evaluated the benefit of teriflunomide at a dose of 7 mg/day or 14 mg/day as monotherapy versus placebo in respect of relapse, disability worsening and safety over at least 48 weeks (a maximum of 173 weeks) (Confavreux 2014) or two years (O'Connor 2011). The other three studies mainly evaluated the efficacy on relapse and the safety and tolerability of teriflunomide 7 mg/day or 14 mg/day with add‐on IFNβ versus placebo over 48 weeks (Freedman 2012) or at least 24 weeks (a maximum of 108 weeks) (NCT01252355), or teriflunomide 7 mg/day or 14 mg/day alone versus IFNβ‐1a (Vermersch 2014) at least 48 weeks (a maximum of 115 weeks). The study design and interventions of the five studies were diverse. There were obvious clinical heterogeneities due to the diversities in study designs or interventions and methodological heterogeneities across studies. All studies had a high risk of detection bias for relapse assessment and a high risk of bias due to conflicts of interest. Among them, Confavreux 2014, Freedman 2012, and NCT01252355 had a high risk of attrition bias due to a high dropout rate (29.8% (Confavreux 2014), 36.4% (Freedman 2012) and 100% (NCT01252355). O'Connor 2011 and Vermersch 2014 had an unclear risk of attrition bias. Freedman 2012, NCT01252355, and Vermersch 2014had a high risk for performance bias and a lack of power due to the limited sample. The data at one year in O'Connor 2011 were not available. As a result, we could not conduct meta‐analyses. We calculated the treatment effects of interventions based on the available data in the original studies.
Compared to placebo, administration of teriflunomide at a dose of 7 mg/day or 14 mg/day as monotherapy reduced the number of participants with relapse by one year or by two years, as well as the annualized relapse rate by two years. Both doses of teriflunomide as monotherapy reduced the number of gadolinium‐enhancing T1‐weighted lesions by two years, However, only teriflunomide at a dose of 14 mg/day as monotherapy significantly reduced the number of participants with disability progression and delayed the progression of disability by one year and two years. High dose rather than low dose of teriflunomide as monotherapy reduced the volume of hypointense lesions on T1‐weighted images by two years. Neither doses of teriflunomide improved QoL measured by SF‐36 scores by one year. When taking the effect of drop‐outs into consideration, the likely‐case scenario analyses still showed a benefit in reducing the number of participants with relapse, but not for the number of participants with disability progression. When administrated as combination therapy with IFNβ for one year, neither doses of teriflunomide added to IFNβ were superior to placebo added to IFNβ concerning annualized relapse rate and the number of gadolinium‐enhancing T1‐weighted lesions. When compared to IFNβ‐1a, low‐dose teriflunomide was inferior to IFNβ‐1a in respect of the annualized relapse rate and the number of participants with relapse, but there was no difference for high‐dose teriflunomide.
Overall, the risks for AEs and SAEs in participants receiving teriflunomide were similar to those in participants receiving placebo both at one year and two years of follow‐up. However, the risks for study drug discontinuation due to AEs were increased by both doses of teriflunomide administration at one year of follow‐up, but not at two years of follow‐up. The most common
AEs associated with teriflunomide included diarrhoea, nausea, hair thinning, elevated ALT levels, neutropenia and lymphopenia. These AEs rarely led to discontinuation of the study medication, but did have a dose‐related effect.
Overall completeness and applicability of evidence
In this review, we excluded one RCT due to length of follow‐up shorter than one year. Generally, DMT for MS needs an adequate administration duration and follow‐up to determine the benefit and safety outcomes accurately. A minimum duration of administration of one year, pre‐defined in the criteria of types of interventions, was a reasonable treatment length that partly avoided the inclusion of misleading evidence. We did not perform meta‐analyses due to the diversities in the study design and interventions. Two large‐scale RCTs contributed to the main evidence for this review. The evidence was only applicable to adults with relapsing MS, who had a score of less than 5.5 on the EDSS and disease activity with at least one relapse in the previous year or at least two clinical relapses in the previous two years.
Quality of the evidence
We included five RCTs in this review, involving 3231 adults with relapsing MS to evaluate mainly the benefit and safety of two doses of teriflunomide (7 mg/day and 14 mg/day) as monotherapy or combination therapy with IFNβ by direct comparison with placebo or IFNβ‐1a. Overall, there were obvious clinical heterogeneities due to the diversities in study designs or interventions and methodological heterogeneities across studies. All studies had a high risk of detection bias for relapse assessment and a high risk of bias due to conflicts of interest. Among them, three studies also had a high risk of attrition bias due to a high dropout rate and two studies had an unclear risk of attrition bias. Generally, the higher the ratio of participants with missing data to participants with events, the greater potential there is for bias, especially for the high frequency of events. The potential impact of missing continuous outcomes increases with the proportion of participants with missing data. In addition, the studies of combination therapy with IFNβ and the study with IFNβ‐1a as controls also had a high risk of performance bias and a lack of power due to the limited sample. The evidence in this review was mainly derived from the two large‐scale RCTs, in which the high risk of bias lead to low‐quality evidence for the results of relapse. The results of disability progression were also subjected to a serious indirectness of evidence because disability progression was confirmed in less than six months in both studies. The evidence for disability progression was very low.
Potential biases in the review process
An extensive and comprehensive search was undertaken to limit bias in the review process. The two review authors' independent assessments of the eligibility of studies for inclusion in this review and extraction of data minimized the potential for additional bias beyond that detailed in the 'Risk of bias' tables. The review authors had no conflicts of interest.
Agreements and disagreements with other studies or reviews
There are three network meta‐analyses that compared the benefit or acceptability of teriflunomide and other DMDs by mixed treatment comparison (Hadjigeorgiou 2013; Tramacere 2015; Zagmutt 2015). We have found no other systematic review on teriflunomide for MS that used only direct head‐to‐head comparisons. In this systematic review, our extensive and comprehensive search found only one RCT comparing teriflunomide to other DMDs, and it was very low‐quality. There was a lack of evidence to show the comparative benefit and safety of teriflunomide in comparison with other DMDs. As with the results of the network meta‐analyses, this systematic review also emphasized the need for randomized trials of direct comparisons between teriflunomide and other active agents.
Study flow diagram.
Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.
Risk of bias summary: review authors' judgements about each risk of bias item for each included study.
| Teriflunomide compared to placebo for multiple sclerosis | ||||||
| Patient or population: people with relapsing multiple sclerosis | ||||||
| Outcomes | Illustrative comparative risks* (95% CI) | Relative effect | No of Participants | Quality of the evidence | Comments | |
| Assumed risk | Corresponding risk | |||||
| Placebo | Teriflunomide | |||||
| Proportion of participants with at least 1 relapse at 1 year | 394 per 1000 | 237 per 1000 | RR 0.60 | 761 | ⊕⊕⊝⊝ | This outcome was considered low, because we considered there were very serious limitation in study design and execution. The bias that influenced the validity of the results for this outcome included: the high risks of bias due to unblinded assessments for relapse and conflicts of interest (sensitivity analysis according to a likely‐case scenario showed a robustness for the results of this outcome, we considered that the high attrition bias did not influence the robustness of the results on relapse). Therefore, we downgraded the quality of evidence for this outcome by 2 levels |
| The proportion of participants with at least 1 relapse at 2 years | 545 per 1000 | 436 per 1000 | RR 0.80 | 722 | ⊕⊕⊝⊝ | This outcome was considered low, because we considered there were very serious limitation in study design and execution. The bias that influenced the validity of the results for this outcome included: the high risks of bias due to unblinded assessments for relapse and conflicts of interest. (Sensitivity analysis according to a likely‐case scenario showed a robustness for the results of this outcome, we considered that the unclear attrition bias did not influence the robustness of the results on relapse.) Therefore, we downgraded the quality of evidence for this outcome by 2 levels |
| The proportion of participants with disability progression at 1 year | 142 per 1000 | 78 per 1000 | RR 0.55 | 761 | ⊕⊝⊝⊝ | This outcome was considered very low based on the following reasons:
|
| The proportion of participants with disability progression at 2 years | 273 per 1000 | 202 per 1000 | RR 0.74 | 722 | ⊕⊝⊝⊝ | This outcome was considered very low based on the following reasons:
|
| The proportion of participants with diarrhoea at 2 years | 89 per 1000 | 179 per 1000 | RR 2.01 | 718 | ⊕⊕⊕⊝ | The follow‐up periods were diverse in Confavreux 2014 and O'Connor 2011 (at least 48 weeks (Confavreux 2014) and 108 weeks (O'Connor 2011)). Treatment duration of participants in Confavreux 2014 was variable, ending 48 weeks after the last participant was included (a maximum treatment duration of 173 weeks). Actually, the data on adverse events in Confavreux 2014 were not at 2 years. There was a heterogeneity in follow‐up period between the studies. Therefore, we did not combine the data on adverse events in Confavreux 2014 and O'Connor 2011 |
| The proportion of participants with hair thinning at 2 years | 33 per 1000 | 131 per 1000 | RR 3.94 | 718 | ⊕⊕⊕⊝ | The follow‐up periods were diverse in Confavreux 2014 and O'Connor 2011 (at least 48 weeks (Confavreux 2014) and 108 weeks (O'Connor 2011)). Treatment duration of participants in Confavreux 2014 was variable, ending 48 weeks after the last participant was included (a maximum treatment duration of 173 weeks). Actually, the data on adverse events in Confavreux 2014 were not at 2 years. There was a heterogeneity in follow‐up period between the studies. Therefore, we did not combine the data on adverse events in Confavreux 2014 and O'Connor 2011 |
| The proportion of participants with elevated ALT levels at 2 years | 67 per 1000 | 143 per 1000 | RR 2.14 | 718 | ⊕⊕⊕⊝ | The follow‐up periods were diverse in Confavreux 2014 and O'Connor 2011 (at least 48 weeks (Confavreux 2014) and 108 weeks (O'Connor 2011)). Treatment duration of participants in Confavreux 2014 was variable, ending 48 weeks after the last participant was included (a maximum treatment duration of 173 weeks). Actually, the data on adverse events in Confavreux 2014 were not at 2 years. There was a heterogeneity in follow‐up period between the studies. Therefore, we did not combine the data on adverse events in Confavreux 2014 and O'Connor 2011 |
| *The basis for assumed risk is the placebo group risk. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). The assumed risk was defined as placebo group risk because only one study was evaluated. | ||||||
| GRADE Working Group grades of evidence | ||||||
| a High risks of bias existed in Confavreux 2014 due to unblinded assessments for relapse and conflicts of interest. b High risks of bias existed in O'Connor 2011 due to unblinded assessments for relapse and conflicts of interest. c High risks of bias existed in Confavreux 2014 due to effects of the high attrition bias on progression disability and conflicts of interest. d High risk of bias existed in O'Connor 2011 due to effects of the unclear attrition bias on progression disability and conflicts of interest. e Serious indirectness existed in Confavreux 2014 or in O'Connor 2011 because disability progression was confirmed at 3 months of follow‐up. f High risk of bias existed in O'Connor 2011 due to an unclear attrition bias and conflicts of interest. | ||||||
