American College of Physicians diabetes guidelines attempt to turn back the clock, conflating good HbA1c with hypoglycemia

These guidelines are not only regressive, going back to the care of patients with diabetes that was current 10–15 years ago, but also misleading. The authors term the guidelines “evidence-based”, yet limit their evidence to older studies, choosing to ignore the newer studies and therapeutics of the past decade. They also ignore widely available and accepted data proving that the lower the HbA1c, the fewer complications people have.

Why do we consider this to be erroneous, and how should recommendations to improve glycemic control be explained in light of up-to-date approaches?

The ACP group misinterprets a group of cardiovascular outcome studies performed and reported a decade ago in which approaches to glucose lowering caused excess hypoglycemia and led to equivocal evidence of glycemic benefit. The ACP highlights the Action to Control Cardiovascular Risk in Diabetes (ACCORD) Trial, in which 257 deaths occurred over 3.5 years with an achieved HbA1c of 6.4% among 5128 intensively treated high cardiovascular risk patients, compared with 203 deaths with an achieved HbA1c of 7.5% among 5123 patients having “standard treatment”.2 Most importantly, the ACP confuses the benefits of a lower HbA1c goal with the adverse effect of treatments that cause hypoglycemia.

Hypoglycemia emerged as a much stronger predictor of adverse outcome in ACCORD, associated with more than doubling of mortality in the intensively treated group and with more than tripling in the standard treatment group.3 Furthermore, it was not the lower HbA1c target that predicted a worse outcome. Rather, those intensively treated people whose on-trial HbA1c levels were lower actually had a reduction in mortality compared with the standard control group, whereas those with higher on-trial HbA1c had a higher mortality rate. In fact, at the time the study was stopped due to the excess mortality in the intensively managed group, there was already a 10.5% reduction in non-fatal myocardial infarction (MI).4 The increase in mortality was only seen in the subset of people with higher HbA1c levels at baseline.5 The approach taken to intensive treatment in that trial involved more use of bolus insulin, thiazolidinediones, and repaglinide, with more than triple the likelihood of use of three, four, or five classes of glucose-lowering medicines with insulin,6 an approach now recognized to be particularly likely to increase hypoglycemia rates.

An important caveat is that HbA1c only partially reflects mean glucose levels.7 A recent study using continuous glucose monitoring suggested that approximately half the variance in HbA1c is not explained by mean glucose.8 Chronic kidney disease and other states associated with anemia are conditions in which low HbA1c often is discordant with glucose levels.9 The important implication: rather than being a marker of hypoglycemia risk, low HbA1c may track with a variety of illnesses in which adverse cardiovascular outcome and mortality occur at such increased frequency as to obscure the beneficial effect of better glycemic control.

Progression of diabetic retinopathy, surely a significant complication, was reduced by more than half among participants in ACCORD undergoing intensive rather than standard glycemic treatment,10 and allocation to intensive rather than standard glycemic treatment in ACCORD significantly reduced MI, coronary revascularization, and unstable angina, an effect explained by the reduction in HbA1c.11 In the UK Prospective Diabetes Study, the mean HbA1c of 7.9% in the control group (a level that would surely be typical if the ACP guidelines were accepted) led to a 46% rate of diabetes-related endpoints over the course of the study, 12% higher than the 41% rate among those randomized to the intensive intervention with mean HbA1c of 7.0%, with 25%–30% reductions in vision loss and need for laser photocoagulation, a 33% reduction in the development of microalbuminuria, and a 16% reduction in MI,12 with subsequent 10-year follow-up revealing the same reduction.13 In the Action in Diabetes and Vascular disease: Preterax and Diamicron Modified Release Controlled Evaluation (ADVANCE) trial, every 1% increase in HbA1c levels above 7.0% for macrovascular events and death, and above 6.5% for microvascular events, was associated with a 38% higher risk of a macrovascular event, a 40% higher risk of a microvascular event, and a 38% higher risk of death, with all effects highly significant.14

We have argued elsewhere that avoiding hypoglycemia is critical15 and reducing treatment certainly is appropriate in people with diabetes experiencing significant low glucose levels. However, conflating low HbA1c with hypoglycemia constitutes a conceptual and logical error. We presume that the notions espoused by the ACP group are based on this misunderstanding, but would urge that such be avoided. The use of therapies such as the thiazolidinediones, sodium–glucose cotransporter 2 inhibitors, and glucagon-like peptide-1 receptor agonists associated with cardiovascular benefit over and above that due to glycemic control, and not intrinsically causing hypoglycemia, represents important and growing understanding of correct approaches to the treatment of type 2 diabetes (T2D). We look forward to helping the ACP to recognize such approaches in their ongoing efforts to improve outcomes of treatment of people with T2D, and we would emphasize that there is potential for harm from establishing goals well above glycemic levels that have been shown to reduce diabetes complications.

Should we “de-intensify” treatment for those with HbA1c below 6.5%?

This is a particularly bizarre suggestion in a set of “evidence-based recommendations” because there is no evidence, no controlled study, and no “real-life big data” that “de-intensified” treatment improves outcome. The only time that de-intensification improves the patient's condition is when drugs are stopped for hypoglycemia and/or other adverse effects. Why would deliberate measures be taken to worsen the level of glycemia of a person with T2D showing excellent glycemic control?

Collaboration of all groups working to improve diabetes outcomes should be our goal, and we should endeavor to optimally individualize care in a manner not exposing people with diabetes to under-treatment, developing globally acceptable guidelines that will address specific situations and deficiencies of care a variety of populations, including those of both developed and developing countries.

美国内科医师学会(ACP)临床指南委员会最近发布了以下指导声明¹:

临床医生应该致力于在大多数2型糖尿病患者中实现7%至8%的HbA1c水平

并且

临床医生应考虑在HbA1c水平低于6.5%的2型糖尿病患者中减弱药物治疗。

这些指南不仅有所退步,  将糖尿病患者的治疗退回到10-15年前的状态,  而且还有误导性。作者将这些指南称为“循证指南”,  但却将证据限制在较早的研究中,  选择忽略过去十年的最新研究和治疗进展。他们还忽略了广泛可用和广泛接受的数据,  这些数据证明HbA1c越低,  患者的并发症就越少。

为什么我们认为这条推荐是错误的,  应该如何根据最新方法解释改善血糖控制的推荐?

ACP小组错误解读了一组十年前进行并报告的心血管转归研究,  在这些研究中降糖治疗引起过度低血糖并导致血糖益处的证据变得模棱两可。ACP强调了在糖尿病患者中控制心血管风险行动(ACCORD)试验,  在此项试验中5128例强化治疗且HbA1c达到6.4%的高心血管风险患者中,  3.5年内死亡257例,  而在5123例“标准治疗”且HbA1c达到7.5%的患者中,  死亡203例²。重要的是,  ACP混淆了较低HbA1c目标的益处和引起低血糖的治疗不良反应。

在ACCORD试验中,  低血糖成为不良转归更强效的预测因子,  与强化治疗组死亡率增加两倍以上以及标准治疗组死亡率增加三倍以上有关³。此外,  更低的HbA1c目标不能预测更差的转归。相反,  与标准对照组相比,  在试验过程中HbA1c水平较低的强化治疗组的实际死亡率更低,  而在试验过程中HbA1c水平较高的患者死亡率更高。事实上,  在因为强化治疗组死亡率过高而停止研究时,  非致死性心肌梗死(MI)的发生率已经降低了10.5%⁴。死亡率增加仅见于基线时HbA1c水平较高的人群⁵。在此项试验中采取的强化治疗方法包括更多使用餐时胰岛素,  噻唑烷二酮和瑞格列奈,  使用三、四或五种降糖药与胰岛素联合使用的可能性高三倍以上⁶。现在已经认识到这种强化治疗方法特别有可能增加低血糖的发生率。

一个重要的警告是,  HbA1c仅部分反映平均葡萄糖水平⁷。最近一项使用连续血糖监测的研究表明,  大约一半的HbA1c变异不能用平均葡萄糖水平进行解释⁸。在慢性肾病和其他与贫血相关的疾病中,  低HbA1c经常与葡萄糖水平不一致⁹。重要的意义是:低HbA1c不是低血糖风险的标志物,  低HbA1c可以追踪多种疾病,  在这些疾病中不良心血管转归和死亡发生频率增加,  可能会模糊优化血糖控制的有益作用。

在ACCORD试验中,  与标准治疗相比,  接受强化治疗的患者进展为糖尿病视网膜病变这一重要并发症的比例减少一半以上¹⁰,  并且在ACCORD试验中分配到强化而非标准降糖治疗组可以显著减少MI、冠状动脉再通和不稳定心绞痛的发生率,  这也可以用HbA1c降低进行解释¹¹。在英国前瞻性糖尿病研究中,  对照组的平均HbA1c为7.9%(如果ACP指南被接受,  肯定是典型的水平),  在研究过程中糖尿病相关终点的发生率为46%,  比随机分配至强化干预组(41%)高12%,  强化治疗组的平均HbA1c为7.0%,  视力下降并且需要激光光凝术的发生率降低25%-30%,  微量白蛋白尿的发生率降低33%,  MI的发生率降低16%¹²,  随后的10年随访显示相同的降低幅度¹³。在糖尿病和血管疾病行动:Preterax和Diamicron 缓释对照评估(ADVANCE)试验中,  HbA1c水平高于7.0%后每增加1%(大血管事件和死亡),  以及HbA1c水平高于6.5%后每增加1%(微血管事件),  与大血管事件风险相关性增加38% ,  微血管事件风险增加40%,  死亡风险增加38%有关,  所有作用都非常显著¹⁴。

我们在其他场合也争论过,  避免低血糖至关重要,  减少治疗肯定适用于血糖水平显著降低的糖尿病患者。然而,  混淆低HbA1c与低血糖构成了概念和逻辑错误。我们认为ACP小组所支持的观念是基于这种误解,  但是应该敦促避免这种误解。使用噻唑烷二酮类药物、钠-葡萄糖协同转运蛋白2抑制剂和胰高血糖素样肽-1受体激动剂等药物治疗对心血管的益处超过血糖控制本身的益处,  并且本质上不会导致低血糖,  这是对2型糖尿病(T2D)正确治疗方法重要且日益深入的理解。我们期待帮助ACP在持续提高T2D患者治疗转归的过程中认识到这些方法,  并且我们强调使用远高于已被证明可以减少糖尿病并发症的血糖水平目标有可能会造成伤害。

对于HbA1c低于6.5%的患者,  我们是否应该“减弱”治疗?

这是一系列循证推荐中特别奇怪的建议,  因为没有证据,  没有对照研究,  也没有“现实生活中的大数据”支持 “减弱”治疗可以改善转归。唯一使用减弱治疗可以改善患者病情的情况是因为低血糖和/或其他不良反应而停止治疗。为什么要在血糖控制良好的T2D患者中特意采取措施来升高血糖呢?

让所有致力于改善糖尿病转归的团队共同合作应该成为我们的目标,  我们应该努力优化个体化治疗,  不让糖尿病患者治疗不足,  制定全球都可接受的指南,  以解决在多种人群的治疗中遇到的特定情况和不足,  包括发达国家和发展中国家。