摘要： 目的 评价干扰素-β（IFN-β）治疗复发-缓解型多发性硬化的有效性和安全性。方法 检索Cochrane 临床对照试验中心注册库、美国国立医学图书馆、荷兰医学文摘、CINAHL、LILACS、PEDRO、中国生物医学文献数据库、临床试验注册中心和世界卫生组织国际临床试验注册平台（检索截止时间：2014 年6 月）；并通过阅读相关论文参考文献，联系参与IFN-β治疗多发性硬化临床试验的研究者和企业，进一步获取研究信息或未发表的数据。由两名评价人员独立筛选研究、提取研究信息和数据、评价偏倚风险。应用Review Manager 软件（Version 5.3.3）进行Meta 分析，GRADEpro 软件评价研究设计和实施过程中的局限性（偏倚风险）、结果的不一致性和不精确性、证据的间接性和发表偏倚对主体证据质量的影响。结果 共检索相关文献576 篇，阅读标题和摘要后初步筛选出26 项研究；进一步阅读全文后纳入5 项研究（共2129 例复发-缓解型多发性硬化患者：高剂量IFN-β组1076 例、安慰剂组1053 例）。所有纳入的研究均为IFN-β单药治疗且随访时间≥ 1 年的随机双盲安慰剂对照平行临床试验。大多数研究存在方法学局限性，主要缺陷为随访偏倚风险较高，且数据分析未使用意向治疗原则，仅919 例受试者（43.17%）的数据可用于分析随访2 年时的主要结局。Meta 分析显示，IFN-β可轻微减少随访2 年时复发病例数（RR = 0.810，95%CI：0.740 ~ 0.890；P = 0.000）和残疾进展病例数（RR = 0.700，95%CI：0.550 ~ 0.880；P = 0.002）；敏感性分析（最差情况的演示分析）显示，IFN-β治疗无效（RR = 1.110，95%CI：0.730 ~ 1.680，P = 0.620；RR = 1.310，95%CI：0.600 ~ 2.890，P = 0.500）。共1581 例患者（74.26%）的数据可用于分析随访1 年时至少复发1 次的病例数（RR = 0.740，95%CI：0.590 ~ 0.930；P = 0.010），绝对危险降低率为13.24%，需治疗的病例数为8例，表明需要治疗8例患者才可防止1例在第1年内复发。但在年复发率方面，IFN-β治疗无效。IFN-β常导致注射部位局部反应、寒颤、发热、肌肉疼痛、流感样症状、头痛、血清丙氨酸转氨酶和天冬氨酸转氨酶水平升高等不良事件，但并不增加外周血淋巴细胞和中性粒细胞减少、抑郁、自杀行为或自杀观念的发生。结论 高质量证据显示，IFN-β治疗复发-缓解型多发性硬化可轻微降低第1 年内的复发病例数，但超过1 年的疗效尚不能确定。目前尚无足够证据证明IFN-β在减少残疾进展病例数方面的疗效，尚待高质量的随机对照临床试验评价其长期有效性。

关键词: 干扰素β, 多发性硬化, Meta分析

Abstract: Objective  To assess the efficacy and safety of interferon-beta (IFN-β) as monotherapy versus placebo for patients with relapsing-remitting multiple sclerosis (RRMS).  Methods  We searched Cochrane Central Register of Controlled Trials (CENTRAL), PubMed, EMBASE, CINAHL, LILACS, PEDRO, China Biology Medicine Disc (CBMDisc), as well as clinical trial registries and the World Health Organization International Clinical Trials Registry Platform (WHO ICTRP, retrieval deadline: June 2014). Furthermore, we checked reference lists of published reviews and retrieved articles, and communicated personally with investigators and biotechnology companies participating in trials of IFN-β in an effort to identify further studies or unpublished data. Two review authors independently screened studies, extracted data and evaluated the risk of bias. Formal Meta-analysis were conducted by using Review Manager software (Version 5.3.3) and the impacts of limitations in study design or execution (risk of bias), inconsistency in results, imprecision of results, indirectness of evidence and publication bias on the quality of the body of evidence were assessed.  Results  A total of 576 articles were retrieved. After screening of titles and abstracts, 26 studies were provisionally selected. The full text of papers were obtained for further assessment of eligibility. Finally, 5 studies were included, involving 2129 patients with RRMS (high-dose IFN-β group: N = 1076; placebo group: N = 1053). All studies were randomized, double-blind, controlled, parallel-group clinical trials with a follow-up for at least one year, evaluating IFN-β versus placebo as monotherapy for patients with RRMS. Most studies had methodological limitations, mainly on a high risk of attrition bias. Moreover, the intention to treat (ITT) principle was not used in data analysis. Data from only 919 patients (43.17%) were available to calculate the primary outcomes at 2 years of follow-up. Meta-analysis indicated IFN-β slightly reduced the number of patients with at least one relapse [risk ratio (RR) = 0.810, 95%CI: 0.740-0.890; P = 0.000] and the number of patients with disability progression during the first 2 years of follow-up (RR = 0.700, 95%CI: 0.550-0.880; P = 0.002). However, the sensitivity analysis (worst-case scenario analysis) showed no treatment effect (RR = 1.110, 95%CI: 0.730-1.680, P = 0.620; RR = 1.310, 95%CI: 0.600-2.890, P = 0.500, respectively). Data from 1581 patients (74.26%) were available to analyze the number of patients with at least one relapse during the first year of follow-up (RR = 0.740, 95% CI: 0.590-0.930; P = 0.010). Absolute risk reduction (ARR) was 13.24% and number needed to treat (NNT) was 8, which meant 8 patients were needed to treat to prevent one patient against relapse. However, the pooled results showed no treatment effect on the annualized relapse rate. The adverse events frequently caused by IFN-β included injection-site reactions, chills, pyrexia, myalgia, influenza-like symptoms, headache, increased alanine aminotransferase (ALT) and increased aspartate aminotransferase (AST). However, the incidences of lymphocytopenia, leucopenia, depression and committed or attempted suicide were not significantly increased by IFN-β.  Conclusions  There is high-quality evidence to support that IFN-β slightly reduces the number of patients with RRMS having relapse during the first year of follow-up, but the clinical effect beyond one year is uncertain. There is insufficient evidence to determine the efficacy of IFN-β in reducing the number of patients with disability progression. New randomized controlled trials of high quality are needed to assess the long-term efficacy.

Key words: Interferon-beta, Multiple sclerosis, Meta-analysis