Cardiovascular safety trials: Be careful what you wish for (心血管安全性研究:当心你许的愿)


All clinicians want medications used for diabetes to be safe. We are dismayed by the not-infrequent occurrence of severe hypoglycemia with the very treatments that we are using to control the disease. Insulin and oral hypoglycemic agents (mainly sulfonylureas) are responsible for more than one-quarter of adverse drug reaction hospitalizations in the US.[1] All the more important, then, that glucose-lowering agents not promote cardiovascular (CV) disease (CVD), explaining the great furor over the supposed increase in CV events with rosiglitazone described by Nissen and Wolski in 2007.[2] Although we[3] and others[4] questioned that meta-analysis, it led to the 2008 US Food and Drug Administration (FDA) requirement that all glucose-lowering agents for which Phase 3 trials cannot exclude a chance of 5% or more that there may be a 1.8-fold increase in CV event rates should undergo a large CV safety trial to exclude a chance of 5% or more that there may be even a 1.3-fold increase in event rates.[5] A short-term study performed in a population with relatively low event rates may not exclude this 1.8-fold increase. Indeed, we have observed that a drug given for 1 year to 2000 patients and 2000 controls, with a 1% event rate in both groups, would have an odds ratio for events of 1.0, but with a 95% confidence interval of 0.5397–1.8528,[6] the upper limit then automatically triggering a CV safety trial.


To make trial sizes manageable, CV safety studies typically recruit individuals whose annual CV risk is far greater than 1%. With greater numbers of events, the confidence limits go down rapidly. This brings us to two recently reported trials of the dipeptidyl peptidase (DPP)-4 inhibitors saxagliptin and alogliptin.


In the Saxagliptin Assessment of Vascular Outcomes Recorded in Patients with Diabetes Mellitus (SAVOR) trial, 8280 type 2 diabetic patients were randomized to saxagliptin and 8212 to placebo.[7] Of the study population as a whole, 79% had evidence of atherosclerotic disease, 82% had hypertension, 71% had dyslipidemia, 38% had prior myocardial infarction, 13% had prior heart failure, and 43% had prior coronary revascularization. The mean age of the study group was 65 years, mean body mass index was 31 kg/m2, and the subjects had had diabetes for 10 years. In that study, 70% of subjects were treated with metformin, 40% with a sulfonylurea, and 41% with insulin. The primary endpoint of CV death, non-fatal myocardial infarction, or non-fatal stroke occurred at annual rates of 3.7% in both groups, whereas an endpoint expanded to include unstable angina hospitalization, coronary revascularization, and heart failure occurred at annual rates of 6.6% and 6.5% in those randomized to saxagliptin and placebo, respectively, demonstrating no significant difference, although 3.5% of those receiving saxagliptin and 2.8% of those receiving placebo were hospitalized for heart failure during the median 2.1-year period of observation. There was an improvement in glycemia, with HbA1c decreasing from 8% to 7.7% at the end of treatment with saxagliptin and to 7.9% with placebo, but hypoglycemia occurred more often with saxagliptin (in 15% vs 13% of subjects), although not with a significant increase in hospitalizations for this. Renal abnormality (not defined explicitly in the report) occurred significantly more often with saxagliptin, although the patients receiving saxagliptin also had a significantly greater likelihood of improvement in the level of albuminuria.

在评估沙格列汀与糖尿病患者中记录到的血管结果关系的试验(Saxagliptin Assessment of Vascular Outcomes Recorded in Patients with Diabetes Mellitus,SAVOR)中,随机将8280名2型糖尿病患者分入沙格列汀组,8212名患者分入安慰剂组[7]。在整个研究人群中,79%的患者被证实有动脉粥样硬化性疾病,82%有高血压,71%有血脂异常,38%既往有心肌梗塞,13%既往有心力衰竭,43%曾行冠状动脉重建术。研究人群的平均年龄为65岁,平均体重指数为31 kg/m2,受试者已患糖尿病10年。在这项研究中,70%的受试者使用二甲双胍治疗,40%使用一种磺脲类药物治疗,41%使用胰岛素治疗。结果显示主要研究终点CV死亡、非致命性心肌梗塞或非致命性中风的年发生率在两组中都是3.7%。扩展终点包括住院治疗的不稳定心绞痛、冠状动脉重建术、以及心力衰竭的年发生率在两组中分别为6.6%与6.5%,没有显著性差异,虽然在中位数为2.1年的观察期中,接受沙格列汀治疗的患者组和接受安慰剂治疗的患者组分别有3.5%和2.8%的人因心力衰竭而住院治疗。血糖也有所改善,在治疗末期,沙格列汀治疗组患者的HbA1c从8%降至7.7%,而安慰剂组患者降至7.9%,但使用沙格列汀治疗的患者低血糖发生率更高(15% vs 13%),虽然因此而住院治疗的患者没有显著增加。使用沙格列汀治疗的患者肾功能异常(在报告中没有明确定义)的发生率明显更高,虽然沙格列汀对患者蛋白尿水平的改善更为显著。

In the Examination of Cardiovascular Outcomes with Alogliptin versus Standard of Care (EXAMINE) study, an even higher risk group was recruited, namely 5380 patients with diabetes who had had recent acute coronary syndrome.[8] The mean age of this study group was 61 years, they had had diabetes for 7 years, 88% had had myocardial infarction, and 28% had congestive heart failure, with 63% and 13% having undergone percutaneous coronary intervention and coronary artery bypass grafting, respectively. Of those receiving alogliptin and placebo, 11.3% and 11.8%, respectively, exhibited the primary endpoint of CV death, non-fatal myocardial infarction, and non-fatal stroke. Baseline treatment included insulin for 30%, metformin for 66%, and a sulfonylurea for 46% of patients, and their baseline HbA1c was also 8%, decreasing by 0.3% with alogliptin but not changing with placebo, with a median 1.5-year period of observation. Hypoglycemia occurred less frequently in that trial, affecting 7% of patients, and at no greater frequency with alogliptin. In subgroup analysis, there was a suggestion that those with diabetes of <5 years duration, those not using insulin, those using metformin, and those with normal renal function or mild impairment may have greater benefit, with diabetes duration of 10 years or more, insulin use, lack of metformin, and moderate or severe renal impairment perhaps associated with better outcome with placebo.

在调查阿格列汀与标准治疗对心血管结果影响的研究(Examination of Cardiovascular Outcomes with Alogliptin versus Standard of Care,EXAMINE)中,所招募人群的风险更高,有5380名近期有急性冠脉综合征的糖尿病患者入组研究[8]。患者平均年龄为61岁,已患糖尿病7年,88%的患者有心肌梗塞,28%有充血性心力衰竭,有63%与13%的患者分别已行经皮冠状动脉介入治疗与冠状动脉旁路移植术。在接受阿格列汀与安慰剂治疗的患者中分别有11.3%与11.8%出现了主要终点事件如CV死亡、非致命性心肌梗塞以及非致命性中风。在基线有30%的患者使用胰岛素治疗,66%的患者使用二甲双胍治疗,46%的患者使用一种磺脲类药物治疗,他们的基线HbA1c也是8%,在中位数为1.5年的观察期中,使用阿格列汀治疗后患者的HbA1c下降了0.3%,但是安慰剂组没有变化。在这项研究中低血糖的发生率较低,只有7%,阿格列汀组的低血糖发生率并没有升高。在亚组分析中发现,糖尿病病程小于5年的患者、没有使用胰岛素治疗的患者、使用二甲双胍治疗的患者以及肾功能正常或轻度受损的患者使用阿格列汀治疗后可能获益更大,而糖尿病病程大于等于10年的患者、使用胰岛素治疗的患者、没有使用二甲双胍治疗的患者以及肾功能中度或严重受损的患者使用安慰剂治疗的结果可能更好。

Should we agree with the initial commentators that these drugs “neither help nor harm CV risk in diabetics”?[9, 10] Two important suggestions of the major CV trials in diabetes have been that glycemic control should begin early in the course of diabetes and that hypoglycemia should be avoided.[11] Both the SAVOR[7] and EXAMINE[8] studies were, by design, performed in patients with long-standing type 2 diabetes and underlying CV complications. It is important to ensure safety in such patients, but it is likely that they are, in fact, the ones least likely to show CV benefit, with the significant interaction of diabetes duration with treatment effect in EXAMINE offering further support of this hypothesis. We do not yet have analyses of either trial to ascertain the effect of hypoglycemia, and certainly await such information. Furthermore, we should recognize that individuals treated with sulfonylureas, given to 40% or more of the participants in the two studies,[7, 8] fail to demonstrate the glucose-dependent glucose-lowering effect of DPP-4 inhibitors, rather experiencing an increase in hypoglycemia as well as weight gain.

我们应该赞同最初人们对这些药物“对糖尿病患者的CV风险既无裨益也无害处”的评价吗[9, 10]?在主要的糖尿病CV试验中有两个重要建议,即在糖尿病早期就应该进行血糖控制以及应该避免低血糖[11]。根据设计,SAVOR[7]与EXAMINE[8]研究都是在病程较长且有潜在CV并发症风险的2型糖尿病患者中进行的。在这样的患者中确保安全性是非常重要的,但实际上他们却可能是最不容易显示CV获益的人群,在EXAMINE研究中,糖尿病病程与治疗效果之间具有显著的相互作用,这进一步地支持了这种假说。我们还没有开始分析且明确两个试验中低血糖所造成的影响,但我们肯定要等待此类信息的到来。此外,我们还应该认识到,使用磺脲治疗的患者在这两个研究中所占的比例为40%或更多[7, 8],在这些患者中没能证实DPP-4抑制剂具有葡萄糖依赖性的降糖作用,但却发现体重增加和低血糖发生率的增加。

Many cultures recognize as folk wisdom that sometimes we realize unintended consequences of our desires. In English, “Be careful what you wish for, you're apt to get it”; in Chinese, “当心你许的愿,说不定真的会变成现实。”. The requirement to undertake CV safety studies may, at best, lead to the impression that diabetes treatment cannot improve CVD.

许多文化都认同一种民间智慧,亦即有时当我们的愿望实现后却出现了意想不到的后果。在英语中,“Be careful what you wish for, you’re apt to get it”;用中文来说就是“担心你许的愿,说不定真的会变成现实”。在最好的情况下,进行CV安全性研究的要求可能会导致人们产生治疗糖尿病不会改善CVD的印象。

The latest victim of this misguided FDA directive to somehow “guarantee” CV safety prior to approval (of necessity such short-term studies would involve relatively few patients at the highest risk for CV events) is lixisenatide (a glucagon-like peptide-1 [GLP-1] receptor agonist). This agent has been available in many countries since February 2013, but its manufacturer withdrew its application to the FDA citing concern about potentially compromising, during the review process, the integrity of CV data being gathered in the ongoing Evaluation of LIXisenatide in Acute coronary syndrome (ELIXA) study.[12] And there may be an even greater concern. Recently, the US FDA declined to license insulin degludec and, rather than following the recommendations of its advisory panel that a CV safety study be undertaken as a post-marketing trial, instead required such a study prior to approval.[13] What if, in the interest of obtaining speedy approval, a trial is carried out in people having high CV risk, as was done in the studies of saxagliptin and alogliptin? Insulin certainly has much greater likelihood of causing hypoglycemia than do the DPP-4 inhibitors, and earlier trials with degludec[14, 15] have used a treat-to-target approach to attain normal (and even low-normal) fasting blood glucose levels. We must be sure to recognize the particularly detrimental associations of adverse CV outcome with hypoglycemia[16] and not, in an effort to assess a new product's CV safety, inadvertently subject study participants to a very real potential for harm.

在新药被批准上市之前不知何故必须“保证”CV安全性(这种短期研究必然将纳入处于最高CV事件风险之中的相对少数的患者),利西拉肽(一种胰高血糖素样肽-1[GLP-1]受体激动剂)是这条被误导的FDA法令的最新受害者。从2013年2月以后,这个药物在许多国家都上市了,但是它的制造厂家却撤回了向FDA所作的申请,原因是担心它可能存在的潜在危害,在审查过程中,所收集到的完整CV数据来自于正在进行的利西拉肽对急性冠脉综合征影响的评估研究(Evaluation of LIXisenatide in Acute coronary syndrome,ELIXA)[12]。甚至还有一个更大的担忧。最近,美国FDA拒绝向德谷胰岛素颁发上市许可证,他们并没有听从诺和诺德公司专家顾问小组的建议,将CV安全性试验作为上市后试验来进行,反而要求在批准上市之前完成这样的试验[13]。如果为了获得快速批准,而在高CV风险的人群中进行试验,就像沙格列汀以及阿格列汀所做的研究一样,结果将会怎样呢?胰岛素治疗导致的低血糖可能性一定比DPP-4抑制剂大得多,并且德谷胰岛素早期试验[14, 15]使用的方法是达标治疗使空腹血糖达到正常(甚至低于正常)。我们必须认识到的是低血糖与不良CV结果之间的不利关系[16],并且要努力避免在评估一种新产品的CV安全性的时候,不经意间又将研究参与者置于一种真正的潜在危害之中。