<International Circulation>: And you do that with a meta-analysis? Do you find a trend among persons with a particular disease as a starting point?
　　Prof. Dominiczak: There have been many meta-analyses. For example， C-reactive proteins (CRP) have more meta-analyses than most of us have had hot dinners. But for these novel biomarkers， you first need to go fishing. You do a hypothesis-free search; I call it a fishing expedition. Sometimes you come up with very exciting things. There seems to be a trans-Atlantic debate in which European colleagues including myself believe that CRP should not be part of any guidelines because we have evidence to the contrary. The majority of European Guidelines do not include CRP as something you have to measure to predict cardiovascular risk. Whereas the majority on North American Guidelines still recommend CRP. We think that the evidence is overwhelming not to and there have been a number of studies over the last few years that showed using pretty powerful meta-analyses and powerful genomic techniques (Mendelian randomization) that CRP is just a bystander and tells you that some inflammatory process is going on but in itself this particular gene protein is not causative in any shape or form. So we wouldn’t put it into guidelines for that reason; because it is not a good predictor. But there are others that come out from the sidelines and one of those is the interleukin-6 receptor variant. It is a simple single nucleotide polymorphism in the human genome， one letter difference， and the amino acid changes therefore the protein behaves like the receptor is blocked. What that does is protect against inflammation， so CRP goes down and fibrinogen goes down， albumin goes up; the picture is that of inhibition of inflammation. So it is perhaps a protective cardiovascular marker rather than the marker turning things bad， the marker is doing good so it is an example of the situation where a biomarker could be the start of new drug development or a new molecular target. We have positive evidence from Mendelian randomization from genomics with large meta-analyzed groups of patients and that was just published this year in The Lancet. There can be good things that come completely out of the blue from the sidelines of thinking about biomarkers. If we compare the biomarkers we have got and try to add them to the established Framingham equation they need to have an added value to be useful. There are a number of ways of looking at that and the ROC curves where we look at the area under the curve， the so-called c-statistic， is one of the ways to do it. Again we see that there are markers such as high-sensitivity troponin， the cardiac marker， or NT-proBNP， another cardiac marker that perform quite well with added value on top of the Framingham equation. But CRP doesn’t which is another piece of evidence as to where to put our money and where to study further developments. In the future， we really need to do trials; to use biomarkers or a panel of new biomarkers and then treat people based on the information from biomarkers and see whether that provides added value and that we prevent better using the biomarkers. But that hasn’t happened yet and is the big job to do in the future. At the moment we are just doing more meta-analyses with more patients and lower p values but we still don’t whether this is something that will change clinical practice. That still remains to be done. What we really want to do in the labs however is to discover something completely new. For this， in our labs in Glasgow， we have been utilizing new area of proteomics where you can look at hundreds of peptides and proteins at the same time in a very simple urine sample. That has been interesting. You can do proteomics on many samples; you can use tissue samples， plasma， serum， but we felt and have evidence that urine is a pretty good sample to study because proteases that eat up and destroy proteins don’t work in urine. What you measure using modern technologies is thousands of peptides very quickly and it is not going to degrade any more so the issue of preventing degradation of a sample is less important because the enzyme that degrades proteins is not in this biofluid. It is also particularly good for everything that has to do with the vascular system because the kidney will filter the small peptides that， for example， are part of the vessel structure and once it has been filtered it is there unchanged. So it is like a mirror of what is happening in the kidney vasculature and maybe all vasculature. The technology that we have established and use is capillary electrophoresis which is very tightly coupled to mass spectrometry and is truly high throughput so thousands of proteins and peptides in an hour or less. It is fast， robust， still a bit expensive (too expensive for clinical use tomorrow) but highly reproducible. With all of these biomarkers， to be able to show the same thing in the same subject or group of subjects at any time is hugely important. Reproducibility is the key. This is the type of pattern that we get; so this is several， but could be up to hundreds of biomarkers that are together on one of these results sets and there is the possibility to use bioinformatic analyses to summarize all of them with one number. We can also pick out components and sequence them and go back to the proteins from where these polypeptides originate but probably the exciting thing is to be able to describe them all together. The issue is not one marker in isolation but the pattern. This is quite appealing because medicine is really a pattern of recognition. It shows a characterization of a group of polypeptides of the right size for this technology to pick up which probably， we hope， describes the extracellular matrix in the changing remodeling vessel. That is what we dream of and there is some evidence that that is the case. We would use large groups of patients (as large as possible for this) with a lot of controls and a lot of dynamic change in the system. The beauty of proteomics is that it is dynamic. With genes， we are born with the genes and we die with the same genes with a little bit of methylation through the lifetime but there isn’t a huge change. In proteomics， everything is dynamic and always changing so intuitively this is good for medicine because in medicine we are really interested in change， progression and regression. That is what we have been doing with proteomics. We have been looking at what a typical drug treatment does. We have been looking at whether short-term or long-term drug treatment makes a difference. And we have been looking at exercise as a dynamic modality which is at still at proof of concept but some interesting stuff is coming from that. It is very important to have hundreds of control samples to make absolutely sure that red herrings are not reported. These are the classic ROC curves that show specificity and sensitivity. These are the types of proteins and peptides that we have identified. They are frequently collagen types which are the building blocks of the vessel walls and the rest would be extracellular matrix proteins that are also important in how and why the vessel behaves. We think that as the disease progresses or regresses during treatment or exercise that there is a dynamic change we hope we can pick up. This， for example， is treatment with angiotensin receptor blockers and short-term treatment (10 weeks) does nothing as indicated by this score which is the bioinformatic result that describes the whole pathway as one number. If you treat for two years， the numbers are still small but if you do treat for two years you improve that situation towards normality; something happens dynamically. The same happens with exercise. These are people with severe coronary artery disease frequently three vessels with a lot of atheroma. Some of them are able to mount high levels of physical activity monitored post-MI and those who manage to exercise well for three months change their classification factor towards normality. This has been done in Glasgow on small numbers of subjects but the whole patterns are reproducible now in four countries in a multi-center study of more than 600 subjects using coronary angiography as the gold standard. To summarize the fishing expeditions， we think we can use this urinary proteomics to look for completely new biomarkers we have never heard of before. We think we can show these dynamic changes. We may find new molecules which could lead to completely new pathways or new molecular targets for either drug discovery or other interventions. At the end of the day， I think we will all be partaking ultimately in something we call systems medicine. The idea is rather than doing one biomarker in one system in one pathway and putting an enormous amount of work into just one， we need to use all modern technology (which to me is genomics， proteomics and metabolomics) like layers of a cake and combine it all together because we humans are very complicated. To understand the complex mechanisms， it would be nice to combine all of this information from modern technology at cellular， organ levels and then at the whole human and disease level. To do this we need more fantastic bioinformatics and we are developing this across the world and certainly in my own University. Communication of data is important but in order to publish things， we now deposit lots of data into public repositories of data and I don’t think we are using that enough. There is now genome-wide association studies (GWAS) data from probably a million individuals worldwide each with millions of SNiPs. Our young mathematically-able colleagues need to work on this so as to be able to join this information to what we are learning everyday in the lab to provide a huge resource to do things better. But at the end of the day， we still need to go back to the lab to refine and validate animal models， in cells and in organs because if we don’t understand the mechanisms， no biomarker is going to help us. We need to understand the disease mechanisms fully and completely before we can make a difference. We have a new European consortium with many colleagues and groups around Europe who are focusing on this biomarkers discovery and this is called EU-MASCARA (Markers for Sub-Clinical Cardiovascular Risk Assessment). This started in the last few months and my colleagues at the University of Glasgow have been working very hard.

　　《国际循环》：你做了一些荟萃分析吗？是否发现某一特定疾病在开始时有一种趋势？
　　Dominiczak教授：已经有很多荟萃分析了。但是，新型生物标志物首先需要寻找。你要做一个没有假设的研究；我把这叫做撒网。有时你会发现激动人心的事情。现在似乎有一场跨越大洋的争论，欧洲专家们认为CRP不应当被收入指南，包括我也是这么认为的，因为我们得出了相反的证据。大多数欧洲指南并没有推荐预测心血管风险的时候必须要测定CRP。而绝大多数北美指南仍然推荐测定CRP。我们认为，不测定CRP的证据是相当多的。在过去的几年里，已经有一系列的研究应用了非常有力的荟萃分析方法和有力的基因组学工具（孟德尔随机化），证实CRP只是个旁观者，提示我们有炎症过程正在进行当中，但是无论从哪方面来讲，CRP蛋白都不是致病因素。鉴于上述原因，我们不应当将CRP收入指南，因为CRP并不是一个好的预测因子。
　　但是，我们还额外发现了一些指标，其中包括IL-6受体突变体。这是人类基因组当中的一个简单的单核苷酸多态性，只有一个碱基的差异，使得氨基酸发生了改变，因此蛋白的作用就像受体被阻滞了一样。这具有抗炎作用，因此CRP水平下降，纤维蛋白原水平下降，白蛋白水平升高，整体上表现为抑制炎症反应。因此IL-6受体突变体是有益的，而不是有害的。这一标志物是有益处的，此种情况下生物标志物可以成为新药研发或一个新的分子靶点的开端。通过针对患者的大型荟萃分析，我们得出了基因组学孟德尔随机的阳性证据，这一研究结果发表于今年的《柳叶刀》杂志上。在思考生物标志物的问题时，可能会有一些令人完全意外的收获。如果我们比较已经被发现的生物标志物，尝试把这些生物标志物加入到已经确定的弗雷明汉风险公式中去，这些生物标志物应当有额外的价值，是有用的。有很多方法可以观察这一点，曲线下面积（所谓的c统计量）就是方法之一。另外，我们看到还有心脏标志物高敏肌钙蛋白或氨基端前利钠尿肽被加入弗雷明汉风险公式时带来了额外的价值。但是，CRP则没有额外价值，因此这再次指示我们应当把钱花到哪里，应当研究哪方面的进展。
　　在未来，我们的确需要开展临床试验；应用生物标志物或一组新的生物标志物，然后根据生物标志物的信息来治疗患者，观察生物标志物是否会带来额外的价值，我们应用生物标志物能够更好地预防疾病。但是，这样的事情还没有发生。这是在未来要做的一件大事。当前，我们只是开展样本量更大的、P值更低的、更多的荟萃分析，但是我们不知道这种做法是否能够改变临床实践。还有一些事情需要完成。
　　我们真正想在实验室中做的事情是发现全新的生物标志物。为了达成这一目标，我们在格拉斯哥大学的实验室应用了蛋白组学这一新技术，可以同时观察极其简便易得的尿液标本中的多肽和蛋白质。这很有趣。对很多样本都可以进行蛋白质组学研究；包括组织标本、血浆和血清。但是，我们有证据显示尿液是一个相当不错的研究样本，因为破坏蛋白质的蛋白酶在尿液中不发挥作用。采用现代技术，我们可以快速测定大量的肽，肽不会降解，因此预防样本降解的问题就不那么重要了，因为尿液中没有降解蛋白质的酶类。另外，尿液尤其适合研究与血管系统有关的多肽，因为肾脏会滤过小分子肽，例如这些小分子肽是血管结构的组成部分，一旦被肾脏滤过，小分子肽就不会发生变化。因此，尿液反映了肾脏血管床的情况，甚至反映了全部血管床的情况。
　　我们已经建立和使用的技术是毛细管电泳，与质谱仪紧密相连，是高通量的，因此在一个小时甚至更短时间内能够分析大量的蛋白质和肽。分析速度快、有力，但是花费还是有点儿高（未来临床应用价格太高了），但是可重复性高。对于生物标志物，能够在同一个受试者或同一组受试者重复结果是非常重要的。可重复性是关键。我们得到了生物标志物的图谱；这包括了几个生物标志物，但也可能结果中包括了大量的生物标志物，有可能利用生物信息学分析方法用一个数字来概括所有的生物标志物。我们也可以取出其中部分生物标志物，通过测序由多肽推测出其原始的蛋白。但是，把所有生物标志物一起描述可能是令人兴奋的。问题不是一个单个的生物标志物，而是生物标志物的图谱。这非常有吸引力，因为医学确实是识别图谱的。我们可以显示大小合适的一组多肽的图谱，希望采用技术我们可以找出那些能够描述变化了的重构血管的细胞外基质。这就是我们的梦想。也有些证据显示就是如此。我们可以纳入大量的患者（越多越好），还有很多的对照，观察系统的大量动态变化。蛋白质组学的美好之处就是它是动态的。
　　就基因来说，人生来就是有基因的，死的时候基因还是那样，只是在一生当中发生了少量的甲基化，但是并没有什么巨大的变化。在蛋白质组学里，一切都是动态的，处于不断的变化之中，因此我们直觉地认为这对医学有好处，因为在医学上，我们确实对变化、进展和逆转感兴趣。这些也是我们应用蛋白质组学技术要观察的。我们观察了一种典型的药物治疗会带来哪些变化。我们观察了短期和长期药物治疗是否有差别。我们观察了运动这种动态模式，目前仍处在概念验证的阶段，但是的确发现了一些有意思的结果。设有大量的对照样本是相当重要的，因为要确信不会报告不相关的结果。这里有显示敏感性和特异性的经典的ROC曲线。这里是我们已经鉴别出来的蛋白质和肽的类型。这些蛋白质和肽通常是胶原蛋白，胶原蛋白是血管壁的组成成分。其余的蛋白是细胞外基质蛋白，这些蛋白对于血管如何作用和血管作用的机制是重要的。我认为，随着在治疗和运动中疾病进展或逆转，我们希望能够捕捉到动态变化。例如，在应用ARB短期治疗时（10周）并没有引起评分的变化，该评分是通过生物信息学结果将整个通路以一个数字表示。如果应用ARB治疗2年时，数值仍然较小，但是情况得到改善，朝着正常的方向发展；有一些动态变化。
　　运动也是如此。比如说重度冠心病患者，他们通常有三支病变，有很多粥样硬化斑块。有些患者在发生心梗后被监测到进行了大量了的体育锻炼，那些努力锻炼3个月的患者其分类因子趋向于正常。我们在格拉斯哥大学实验室在小样本的患者中开展了这一观察，但是整个图谱在纳入600个以上受试者的、4个国家开展的多中心研究中得到了验证，研究以冠脉血管造影作为金标准。
　　概况一下撒网，我认为可以应用尿液的蛋白质组学来寻找以前我们从未听说过的全新生物标志物。我认为我们能够显示出生物标志物的动态变化。我们可能会发现新型分子，从而发现全新的信号转导通路或药物治疗或其他干预的新型分子靶点。最终，我想我们会从事系统医学。其做法不是观察一个通路的一个系统里的一个生物标志物，把大量的工作整合在一起，我们需要应用所有的现代技术（对我来说是基因组学、蛋白质组学和代谢组学），就像一层层的蛋糕一样，我们要把他们整合起来，因为人体太复杂了。为了理解复杂的机制，理想的是能够整合细胞、器官、人体和疾病水平的所有现代技术所得出的信息。为了达成这一点，我们需要非常好的生物信息学技术，全球都在开发这一技术，当然包括我所在的大学。数据的交流是重要的，为了发表文章，我们现在在公共的数据存储处已经存储了大量的数据，我想我们对这些数据的利用还不充分。现在我们有全基因组关联分析 （GWAS）的数据，来自全球大概一亿人，每个患者都有大量的单核苷酸多态性检测结果。我们年轻的、擅长数学的同事们需要就此展开工作，以便把这些信息添加到我们在实验室当中每天所得到的信息当中去，为把事情做得更好提供一个巨大的信息库。但是，最后我们还得回到实验室中在动物模型、细胞和器官中去改进和验证发现的生物标志物，因为如果我们不了解机制的话，没有生物标志物能够帮到我们。我们需要全面了解疾病的发病机制，这样才能有所不同。我们新成立了一个欧洲协助组EU-MASCARA（亚临床心血管风险评估标志物），有很多来自欧洲、专注于发现生物标志物的很多同事和研究小组。该研究组在几个月前成立，我在格拉斯哥大学的同事们一直在非常努力的工作着。

上一页  [1]  [2]