Validating a 3rd-generation methylation measure of accelerated aging: DunedinPoAm4x

我们建议验证并向老龄化领域输出一种新的工具来衡量一个人的速度 生物老化：DunedinPoAm4x。该措施将有助于基础衰老研究，也将有助于 测试干预是否减缓了一个人的衰老速度。我们是在NIA的领导下开始这项工作的 R01 AG032282(Belsky等人，PNAS 2015)。在20年的时间里，我们追踪了一个由19个生物标记物组成的小组 同年(1972-73年)出生的1000人的生理功能- 代表达尼丁的研究。我们的目标一直是在相同年龄的人中测量， 根据老年科学理论，生物衰老的变异定义为：渐进性衰退 同时影响多个器官系统。我们跟踪了心血管、新陈代谢、 通过在以下位置重复生物标志物来评估参与者的肺、肾、牙科、肝脏和免疫功能 年龄26岁、32岁、38岁和45岁(94%保留)。这种独一无二的数据集的增长曲线建模 得出了一个老化速度指标，量化了我们队列中的每个参与者的速度或速度 老龄化(Elliott等人的自然老龄化，2021)。这项工作的下一阶段将机器学习应用于 参与者的年龄-45岁全基因组DNA甲基化数据，训练关于衰老速度的算法 指标得出一个名为DunedinPoAm4x(衰老甲基化速度，4波)的分数。这是技术上的 先进意味着一个人的衰老速度可以从一个单一的血液样本中估计出来，而且 该指标可以输出到任何有血液甲基化数据的研究样本。以前我们有过 据报道，对38岁以下的DunedinPoAm初级版本进行了验证检查，结果显示 衰老评分甲基化速度较快的人随后经历了面部晚期 年龄、认知和身体功能下降、慢性病和早期死亡率(Belsky等人 ELife 2020)。我们的目标是建议对DunedinPoAm4x进行系统的样本外验证评估 19个以上的数据集，测试其在老年人、年轻人、种族/民族群体和几个人中的适用性 国家。每个目标还将测试现有的领先甲基化时钟以进行比较。我们最重要的是 假设DunedinPoAm4x将更精确地表征生物衰老 时钟，这样做将带来时钟以外的附加信息价值。除了AIM之外- 具体的出版物，我们计划最终的、综合性的出版物。我们将向基础科学家和 干预研究人员，一种可靠、有效、开放获取的衡量一个人速度有多快的指标 衰老。

We propose to validate and export to the aging field a novel tool to measure a person’s pace of biological aging: DunedinPoAm4x. The measure will be useful for basic aging research and also for testing whether an intervention has slowed a person’s pace of aging. We began this work under NIA R01 AG032282 (Belsky et al PNAS 2015). Over two decades, we tracked a 19-biomarker panel of the physiological functions of 1,000 individuals born the same year (1972-73) in the population- representative Dunedin Study. Our goal has been to measure, in people the same chronological age, variation in biological aging defined per geroscience theory as: gradual, progressive decline simultaneously affecting multiple organ systems. We tracked declines in the cardiovascular, metabolic, pulmonary, renal, dental, hepatic, and immune functions of participants by repeating biomarkers at ages 26, 32, 38 and 45 years (94% retention). Growth-curve modelling of this one-of-a-kind dataset yielded a pace-of-aging metric that quantified how slowly or rapidly each participant in our cohort had been aging (Elliott et al Nature Aging, 2021). The next stage of the work applied machine-learning to participants’ age-45 whole-genome DNA methylation data, training an algorithm on the pace-of-aging metric to derive a score called DunedinPoAm4x (Pace of Aging methylation, 4 waves). This technical advance means that a person’s pace of aging can be estimated from just a single blood sample, and the metric can be exported to any research sample that has blood methylation data. Previously we had reported validation checks on a preliminary version of DunedinPoAm, up to age 38, showing that people who had faster methylation pace of aging scores subsequently experienced advanced facial age, declines in cognitive and physical functioning, chronic diseases, and early mortality (Belsky et al eLife 2020). Our Aims propose a systematic out-of-sample validation evaluation of DunedinPoAm4x in 19+ data sets, testing its applicability in older adults, young people, race/ethnic groups, and several countries. Each Aim will also test existing leading methylation clocks for comparison. Our overarching hypothesis is that DunedinPoAm4x will characterize biological aging with greater precision than the clocks, and in so doing will bring added information value over and above the clocks. In addition to aim- specific publications, we plan a final, synthesis publication. We will deliver, to basic scientists and intervention researchers, a reliable, valid, open-access measure of how rapidly a person has been aging.