“Choices are the hinges of destiny”

—Edwin Markham

Established more than a century ago, and buttressed by 1962 legislation, the United States Food and Drug Administration (FDA) plays a central role in advancing patient care. The FDA’s impressive global reputation has been deservedly earned by providing independent, critically appraised, evidence-based assessments of new therapies before their approval. Public confidence that the FDA product label fairly states the risks and benefits is garnered and maintained by a transparent review process, further supported by the regular establishment of expert external advisory panels on challenging questions related to balancing risk and benefit.

In July 2011, some 20 months after AstraZeneca submitted their new drug application, the FDA granted approval for ticagrelor in patients with acute coronary syndromes based on results from the PLATO (Platelet Inhibition and Patient Outcomes) trial. This prolonged review process was characterized by internal debate related to several issues, including lack of efficacy in the subset of patients enrolled in the United States, as well as a potential negative interaction with aspirin doses >100 mg. Ultimately, an advisory committee convened on July 28, 2010, to address these issues voted 7 to 1 for approval. The final product label contained boxed warnings about potential significant and sometimes fatal bleeding. Additionally, the agency’s risk evaluation and mitigation strategy determined that “Brilinta (ticagrelor) poses a serious and significant public health concern requiring the distribution of a Medication Guide.”

Based on subsequent evidence, ticagrelor received FDA approval, not only for patients with acute coronary syndromes but also for reducing the risk of stent thrombosis, for secondary prevention post–myocardial infarction, and for the treatment of acute ischemic stroke. Most recently, on May 28, 2020, the FDA also approved ticagrelor for the reduction of “the risk of a first [myocardial infarction] or stroke in patients with coronary artery disease at high risk for such events.” This change in the ticagrelor label was based on results from THEMIS (Ticagrelor on Health Outcomes in Diabetes Mellitus Patients Intervention Study). Interestingly, this label modification extended the indication for ticagrelor beyond the trial population studied in THEMIS, which comprised patients with stable coronary disease and type 2 diabetes.¹

In this issue of Circulation, Lackey et al² provide useful insights into how, for the first time, the FDA used “multi-criteria decision analysis” to inform this decision. Their request to an internal Decision Support and Analysis Team was prompted by challenges the FDA faced in assessing whether the benefits observed in THEMIS outweighed the risks. The Decision Support and Analysis Team undertook to “quantitatively explore the sensitivity of the benefit–risk balance to judgements about the relative importance of [6] outcomes: [cardiovascular] death, [myocardial infarction], ischemic stroke, fatal bleeds, intracranial hemorrhage, and other major bleeds.”² Subjective tradeoffs were then used to derive relative weights of the outcomes. Substantial internal disagreement was noted, especially with respect to evaluating weights for stroke and intracranial hemorrhage (see Table by Lackey et al).²

Table. An Approach to Balancing Benefit and Risk in Clinical Trials*

* Adapted from Armstrong and Westerhout with permission. Copyright ©2017, the American Heart Association.⁵

Several points within this process are noteworthy. The new label indications, based on the THEMIS trial results, extend ticagrelor use to prevent myocardial infarction and stroke in patients at high risk for coronary artery disease, yet during a 36-month period, the THEMIS event rates were modest (placebo, 7.6%); by contrast, a similar number of the same events were acquired within approximately 6 months in the truly high-risk patients in the PLATO trial. Balancing the continuous risk of bleeding on a potent antiplatelet therapy in a population at lower risk for ischemic events (where effective alternate therapy exists) is complex. Discarding minor and moderate bleeding from this exercise and suggesting that major bleeding is manageable is a debatable approach. Minor and moderate bleeding episodes may have an important impact on patient quality of life and lead to cessation of antiplatelet therapy. Moreover, there are unintended consequences of interrupting antiplatelet therapy that may be hazardous. Although subgroups were considered as part of the FDA’s review, they were not addressed in the revised label. Many will question this omission given the excess bleeding hazard in women and the favorable net clinical benefit of ticagrelor subsequently reported in the prespecified subgroup with history of percutaneous coronary intervention (58%), in whom previous dual antiplatelet therapy may have pretested a population that subsequently had no excess risk of intracranial bleeding, unlike the non–percutaneous coronary intervention population.³

Interestingly, even though the most unequivocal outcome of cardiovascular death “trended adversely,” the Decision Support and Analysis Team concluded that the true effect was “at worst neutral,” deriving reassurance from the favorable impact on cardiovascular death in the much higher-risk PLATO patient population.² One might equally argue the converse to be true in the lower-risk, stable THEMIS population which experienced less-than-expected benefits. Despite lacking data on the relative gravity between and within the nonfatal outcomes, it was concluded that there was sufficient evidence to support product label modification. Notwithstanding the challenges within the additional internal FDA review, no advisory committee was deemed necessary because THEMIS “met the standard for substantial evidence of effectiveness and there were no scientific or technical matters at issue.”² However, given the number needed to treat of 138 patients to prevent 1 event during 36 months versus 93 patients needed to cause 1 major bleeding event, the THEMIS investigators concluded that ticagrelor therapy did not have a favorable risk–benefit ratio: the accompanying editorialist similarly opined ticagrelor could not be recommended for most trial patients.^1,4

The FDA’s efforts to better estimate the balance between efficacy and safety within composite end points should be applauded given the increasing prevalence of this strategy in clinical trials. The best approach to this vexing issue has previously been addressed in Circulation and elsewhere.⁵ Scaling relative weights of end points in cardiovascular trials to achieve better discernment of their clinical consequences appears to add value to both their interpretation and potential clinical application by providing more complete use of all available data. For example, the major heterogeneity in defining myocardial infarctions in clinical trials stimulated a consensus for a universal definition and a call for consistency in biomarker quantification to facilitate cross-trial comparisons. The broad spectrum of stroke, from transient reversible arm weakness to persisting dense hemiplegia, signals that “weighting within the weights” holds promise in unmasking both benefits and harms of the novel therapy.⁵

Future efforts by the FDA at standardizing these and other key outcome measures, prespecifying their importance, and engaging in a multidisciplinary approach with clinicians, trialists, sponsors, and patients in a more public discussion seems likely to be the preferred path forward (Table). At this crucial hinge point, making more informed choices in approving new therapies will enhance the destiny of our patients.

The author thanks Dr Cynthia Westerhout and Dr Jeffrey Bakal for their reviews and Kris Reay for editorial assistance. The author also acknowledges that some direct quotations regarding the FDA deliberations were acquired from the fda.gov website.

None.

Disclosures Dr Armstrong reports research grants from Merck, Bayer, CSL, Boehringer Ingelheim, and Eli Lilly.

https://www.ahajournals.org/journal/circ

The opinions expressed in this article are not necessarily those of the editors or of the American Heart Association.

For Sources of Funding and Disclosures, see page 584.

“选择是命运的枢纽”

——埃德温·马卡姆

美国食品和药物管理局 (FDA) 成立于一个多世纪以前，并受到 1962 年立法的支持，在推进患者护理方面发挥着核心作用。FDA 令人印象深刻的全球声誉通过在新疗法获得批准之前提供独立的、经过严格评估的、基于证据的评估而当之无愧。公众对 FDA 产品标签公平说明风险和收益的信心是通过透明的审查过程获得和维护的，此外还通过定期建立专家外部咨询小组来解决与平衡风险和收益相关的具有挑战性的问题。

2011 年 7 月，也就是阿斯利康提交新药申请大约 20 个月后，FDA 根据 PLATO（血小板抑制和患者结局）试验的结果批准了替格瑞洛用于急性冠脉综合征患者。这个漫长的审查过程的特点是与几个问题相关的内部辩论，包括在美国招募的患者亚组中缺乏疗效，以及与阿司匹林剂量 >100 毫克的潜在负面相互作用。最终，一个咨询委员会于 2010 年 7 月 28 日召开会议以解决这些问题，以 7 比 1 投票通过。最终产品标签包含有关潜在严重甚至致命出血的黑框警告。此外，

基于随后的证据，替格瑞洛获得了 FDA 的批准，不仅用于急性冠脉综合征患者，还用于降低支架血栓形成的风险、用于心肌梗塞后的二级预防以及用于治疗急性缺血性中风。最近，在 2020 年 5 月 28 日，FDA 还批准替格瑞洛用于降低“具有此类事件高风险的冠状动脉疾病患者首次[心肌梗塞]或中风的风险”。替格瑞洛标签的这一变化基于 THEMIS（替格瑞洛对糖尿病患者干预研究的健康结果）的结果。有趣的是，这一标签修改将替格瑞洛的适应症扩展到 THEMIS 研究的试验人群之外，该人群包括稳定型冠心病和 2 型糖尿病患者。¹

在本期Circulation 中，Lackey 等人²就 FDA 如何首次使用“多标准决策分析”为该决策提供信息提供了有用的见解。FDA 在评估 THEMIS 中观察到的收益是否大于风险时面临的挑战促使他们向内部决策支持和分析团队提出要求。决策支持和分析团队承诺“定量探索利益-风险平衡对判断 [6] 结果的相对重要性的敏感性：[心血管] 死亡、[心肌梗塞]、缺血性中风、致命性出血、颅内出血、和其他大出血。” ²然后使用主观权衡得出结果的相对权重。注意到内部存在重大分歧，特别是在评估中风和颅内出血的权重方面（见拉基等人的表格）。²

桌子。在临床试验中平衡收益和风险的方法*

* 经许可改编自 Armstrong 和 Westerhout。版权所有 ©2017，美国心脏协会。⁵

这个过程中有几个点值得注意。基于 THEMIS 试验结果的新标签适应症扩大了替格瑞洛的使用，以预防冠状动脉疾病高危患者的心肌梗塞和中风，但在 36 个月期间，THEMIS 事件发生率适中（安慰剂，7.6%）；相比之下，PLATO 试验中真正的高危患者在大约 6 个月内获得了相似数量的相同事件。在缺血性事件风险较低的人群（存在有效的替代疗法）中，平衡强效抗血小板治疗的持续出血风险是复杂的。排除此练习中的轻微和中度出血并建议大出血是可以控制的是一种有争议的方法。轻微和中度出血事件可能对患者的生活质量产生重要影响，并导致停止抗血小板治疗。此外，中断抗血小板治疗可能会产生意想不到的后果，这可能是危险的。尽管亚组被认为是 FDA 审查的一部分，修订后的标签中没有提及这些问题。考虑到女性出血风险过多以及替格瑞洛的有利净临床益处随后在有经皮冠状动脉介入治疗史的预定亚组 (58%) 中报告，许多人会质疑这种遗漏，其中先前的双重抗血小板治疗可能已经预先测试了随后的人群与非经皮冠状动脉介入治疗人群不同，没有额外的颅内出血风险。³

有趣的是，尽管心血管死亡最明确的结果“呈不利趋势”，但决策支持和分析团队得出的结论是，真正的影响“在最坏的情况下是中性的”，这从风险高得多的 PLATO 对心血管死亡的有利影响中得到了保证患者群体。²人们可能同样认为，在收益低于预期的低风险、稳定的 THEMIS 人群中，情况正好相反。尽管缺乏关于非致命结果之间和内部的相对严重性的数据，但得出的结论是有足够的证据支持产品标签修改。尽管在额外的 FDA 内部审查中存在挑战，但认为没有必要设立咨询委员会，因为 THEMIS “符合有效性实质性证据的标准，并且不存在科学或技术问题。” ²然而，考虑到在 36 个月内需要治疗 138 名患者以预防 1 次事件，而需要治疗 93 名患者需要引起 1 次大出血事件，THEMIS 研究人员得出结论，替格瑞洛治疗没有有利的风险 - 收益比：随行的编辑类似大多数试验患者不推荐使用替格瑞洛。^1,4

鉴于该策略在临床试验中的流行程度越来越高，FDA 在复合终点内更好地评估疗效和安全性之间的平衡的努力值得称赞。解决这个棘手问题的最佳方法之前已经在Circulation和其他地方解决过。⁵通过更完整地使用所有可用数据，缩放心血管试验中终点的相对权重以更好地识别其临床后果似乎可以增加其解释和潜在临床应用的价值。例如，临床试验中定义心肌梗死的主要异质性激发了对通用定义的共识，并呼吁生物标志物量化的一致性以促进交叉试验比较。广泛的中风，从短暂的可逆手臂无力到持续的密集偏瘫，表明“权重内的权重”有望揭示新疗法的利弊。⁵

FDA 未来在标准化这些和其他关键结果指标、预先指定它们的重要性以及与临床医生、试验者、赞助商和患者进行更公开讨论的多学科方法方面所做的努力似乎可能是首选的前进道路（表）。在这个关键的转折点上，在批准新疗法时做出更明智的选择将改善我们患者的命运。

作者感谢 Cynthia Westerhout 博士和 Jeffrey Bakal 博士的审阅以及 Kris Reay 的编辑协助。作者还承认，有关 FDA 审议的一些直接引述是从 fda.gov 网站获得的。

没有任何。

披露Armstrong 博士报告了默克、拜耳、CSL、勃林格殷格翰和礼来的研究资助。

https://www.ahajournals.org/journal/circ

本文中表达的观点不一定是编辑或美国心脏协会的观点。

有关资金来源和披露信息，请参见第 584 页。