为何?医院抗生素管理策略对阻止抗生素耐药的效果甚微

为何医院的抗生素管理策略对阻止抗生素耐药的效果甚微?


翻译:王宁宁  审核:陈志锦
(SIFIC热点团队)

随着抗生素耐药性的惊人增长,越来越多的医生最后不得不使用抗生素来挽救病人生命,医院是否有更好的方法来管理抗生素治疗方案呢?上一代人之前,循环策略和混合策略这两种抗生素策略被用来战胜细菌。循环策略就如同抗生素“轮作”,即在某一时间段内,某些抗生素交替停止使用。医生认为这种策略能够对抗细菌的耐药性,这是因为“轮作”式使用抗生素是和细菌生活方式相关,从而导致细菌病原体失去抵抗抗生素治疗的能力。混合策略,同样被认为可以减少耐药性,其根源于计算机预测和流行病学模型,这种策略能将抗生素随机分配到患者,在一定合理范围内以最快速度锁定细菌为目标。

根据在《分子生物学与进化》高级在线版发表的一项新研究中进行的分析,事实上这两种策略都并不起作用。由Robert Beardmore、Rafael Pena Miller、Fabio Gori等数学家和生Jon Iredell医开展的理论工作或许能帮助解释在最近“土星项目”的临床试验研究中得出此结论的原因,“土星项目”旨在解决持续争议的抗生素循环策略与混合策略问题。在这个项目中,研究人员得出结论:混合策略和循环策略对抗生素耐药的流行率无统计学显著性差异。

据Beardmore等人的研究,该团队已经证明,“即使利用数学模型进行标准化来确定循环策略或混合策略是否能为耐药微生物选择最佳抗生素都是不可能的,更不用说在临床实践中。”

相反,在《分子生物学与进化》的这项研究中,这支由临床科学家和数学家组成的国际研究团队推荐其他用药策略,如“反应性循环”,他们已经证明,这种策略在所有测试的数学模型中表现均优于循环策略和混合策略。

Beardmore说:“他们的研究结果可能会对未来的临床试验产生深远的影响。从数学角度来说,本次研究的结果非常明确地说明了,抗生素混合策略不是将抗生素分配给患者的最佳方式,而一些临床医师之前却错误地相信了。但是,鉴于数学概念的复杂性,很难将这一点传达到位。最后,真正的数学最优化只代表了一个常识:尽快地把正确的药物使用到正确的患者身上!”

Gory补充道:“以前的研究没有意识到这一点,因为他们过度依赖计算机模拟,而没有全面描述抗生素的优化问题。当我们使用在太空竞赛时期开发的分析技术来解决优化问题 ,一些新的解决方案反而脱离了实际分析。”

该研究团队建议个体化治疗,包括病原体特异性治疗和患者特异性治疗,这些可能是合理优化抗生素使用的其中一个必要条件。他们主张使用通过分子信号技术及血培养的方法术检测出导致感染的病原体,然后通过使用计算机模型来研究不同的个性化医疗方案。

Pena Miller说:“显然,信息量丰富的个性化方案在数学模型中优于抗生素的循环策略和混合策略。而且这一个结论同样适用于细致模型环境。例如,患者进行治疗前已经存在感染,最为糟糕情况是发展为多重耐药菌感染,此时患者能接受的治疗方案就会变得很少。但是,在出现多重耐药菌感染之前,尽可能地采取个体化治疗方案,其效果会优于混合策略及循环策略。”

Added Iredell补充道:“随着临床试验在治疗应用方面的显著增加,个性化用药正在迅速成为现实,严重感染的抗生素使用仍然是医学中最大的难题之一,因此“智能化”使用抗生素将成为优化当下患者疗效和保持长期收益的关键。”

来源:《分子生物学与进化》牛津大学出版社


原文:Why Hospital Antibiotic Management Strategies Do Little to Curb Resistance

With an alarming growth in antibiotic resistance and doctors increasingly having to resort to last-resort antibiotics to save patients, is there a better way for hospitals to manage antibiotic treatment regimens? A generation ago, two antibiotic strategies known as cycling and mixing were employed to outwit bacteria. Cycling is like antibiotic crop rotation where certain classes of antibiotics are withdrawn for a period of time. Doctors thought this would combat resistance because bacterial pathogens would lose their abilities to resist treatment because of the costs associated with a drug-resistant lifestyle. The mixing strategy, with its roots in computer predictions and epidemiological models of the time, was thought to reduce drug resistance because the random assignment of antibiotics to patients, within the appropriate class, would give bacteria the fastest possible moving target.

In reality neither strategy works, according to the analysis performed in a new study published in the advanced online edition of Molecular Biology and Evolution. This theoretical work, by mathematicians Robert Beardmore, Rafael Pena-Miller, Fabio Gori and clinician Jon Iredell, may help explain why recent clinical trials like the Saturn project -- explicitly designed to resolved the ongoing issue of high controversy (antibiotic cycling vs mixing) -- may not work. In the Saturn project, the researchers concluded that there were no statistically significant differences in the prevalence of antibiotic resistance during mixing and cycling interventions.

The team have shown that "determining whether cycling or mixing selects best against drug resistant pathogens is not possible, even in standardized questions using mathematical models, let alone in the clinic," according to Beardmore.

Instead, in the MBE study, the international team of clinical scientists and mathematicians recommends other strategies, like "reactive cycling" which they have shown outperforms cycling and mixing in all the computational models they tested.

Their results could have profound implications for future clinical trials. "Mathematically speaking, it was very clear early in this study that antibiotic mixing was not the optimal way of allocating antibiotic to patients yet this is what some clinicians have come believe," said Beardmore. "But communicating this was difficult, given the complexity of the mathematical ideas. In the end, the real mathematical optimum is little more than common sense: get the right drugs to the right patients as soon a possible."

Gori added, "Prior studies did not see this due to their over-reliance on computer simulations that didn't paint a full picture of the antibiotic optimization problem. When we used an analysis technique developed during the space race era developed to solve optimization problems, some new solutions dropped out of that analysis."

They recommend that individualized treatments, both pathogen-specific and patient-specific, may be a necessity to properly optimize antibiotic use. By using computer models to study different personalized medicine scenarios they advocate for the use of devices that target infections based on rapid diagnoses of the pathogen responsible for the infection from molecular signatures or blood cultures."

"It is clear that information-rich, personalized protocols can outperform antibiotic cycling and mixing in mathematical models but even this conclusion can depend on nuanced model circumstances," said Pena-Miller. "For example, in the doomsday scenario that multi-drug resistance is endemic and present in every infection before the patient begins treatment, it matters little which treatment patients are given. But before that stark situation arises, targeting appropriate treatment at as many individuals as possible outperforms both mixing and cycling."

"Personalized medicine is rapidly becoming a reality with dramatic increases in the availability of clinical testing at the point of care," added Iredell, "Antibiotic use in severe infection remains one of the most powerful interventions in medicine, and intelligent use of antibiotics is essential to optimize immediate patient outcomes and to preserve long-term benefits."

Source: Molecular Biology and Evolution (Oxford University Press)
图文编辑:小小牧童审稿:陈文森 卢先雷


0 个评论

要回复文章请先登录注册