采访嘉宾：

PROFESSOR AMIN ROSTAMI-HODJEGAN, PhD, FCP, FAAPS, FJSSX

Amin is a Professor of Systems Pharmacology and the Director of the Centre for Applied Pharmacokinetic Research (CAPKR) at the University of Manchester. He has an active program of training PhD students involving proteomics, physiologically-based pharmacokinetics and pharmacodynamics, and precision dosing within CAPKR.

Professor Rostami’s scientific work over the last 30 years has covered many aspects of drug development ranging from pharmaceutics (e.g. bioequivalence and formulation) to clinical and systems pharmacology (e.g. mixture pharmacology of parent drug and metabolites and provision of drug-independent system parameters for scaling from in vitro to in vivo studies).

Amin has authored/co-authored over 280 highly cited articles (>16,000 citations, H-Factor = 69). In 2017, he was listed by ISI as one of the world’s most highly cited researchers (under ‘Pharmacology & Toxicology’). He serves on the Editorial Boards of several journals, has been an invited speaker at over 200 international and national meetings and has led numerous workshops in the area of IVIVE-PBPK linked models.

采访人：        

        Yuyan Jin (金玉燕) is the site head of pharmaceutical sciences in China. The function of Pharmaceutical Sciences in China focuses on clinical pharmacology, drug metabolism, toxicology, bioanalytics, and biomarker development to enable drug discovery and development from early to late stages. Yuyan joined Roche in Jan 2013 and have supported global and China development of multiple early and late stage programs. Before joining Roche, she was a Project Clinical Pharmacologist at Pfizer US with a focus on CNS drug development after two-year Critical Path fellowship at US FDA. Yuyan has a BSc In Biopharmaceutics, a MSc in Pharmaceutics from China Pharmaceutical University, and a PhD in quantitative clinical pharmacology from the University of Pittsburgh.

访谈内容：

Dr. Yuyan Jin: You had a presentation in FDA workshop in terms of the open source code and had additional interview to open up the arguments further. Can you briefly share with us your view on this topic?

Professor Amin Rostami-Hodjegan:  I think it would be a good idea if actually I will share my screen. This is from the latest set of the FDA workshop and the presentation that you are referring to.

There are publications favoring the open source code mainly for the following two reasons. The first objection that people have got for the closed systems is a black box. That means that we do not have the trust and that is the reason we want to open them up and understand them. Everybody should be able to see it etc. It is saying that we have to gather trust. The other argument is that if somebody wants to add and modify and so on they are not able to do that if it is not open source code.

However, I brought up in the commentary that published in CPT:PSP is from the angles that there are something missing when they are going one sided. Number one is that when you are trying to submit something to the regulators or you want to make a very big decision, you have to trust that the results of the modeling and simulation is repeatable. It can be trusted with the view that no one has interfered with any of the elements without other people knowing. Open source code is somewhat vulnerable to this element because anybody can go and change some elements of it. Therefore, what I tried to explain in this particular commentary was that we could go away from the black box approach, which nobody can see, and create something that I called glass box. The glass box means that the elements are known but only certain people can change it.

I argued also in favor of open source code as an academic. My students would like to play with all different elements. We also particularly need open source code for academic research. But the bottom line of this is basically to find a balance between what we call open source code: everything is available to everybody versus a closed form that not everyone can touch it. Certain people can look at it, and of course regulators or whoever wants to assess it should have access to and understand it, but at the same time, the glass box have got rid of this vulnerability of the open system.

Dr. Yuyan Jin:  And I have another related question: can you share with us your view on a new term called “model master file”?

Professor Amin Rostami-Hodjegan:  This was the title of the talk that was again in the recent FDA workshop. Dr. Liang Zhao gave their presentation before me that reflects some thinking in the FDA. I hope that the presentation will become available soon. Model Master File contains sets of information that not everybody may know, but now I put it in the form of stopping us from starting from scratch, which is a major problem as we get into complex models.

If you are borrowing a lawn mower from your neighbor. Instead of giving you the lawn mower, imagine they gave you different pieces of it, saying that you make your own lawn mower. This is unnecessary and obviously, based on time they can help you by giving you the pieces but also giving you a map. However, imagine the map is not actually matching. The ideal situation is just to get the lawn mowed. That does not mean that the person who was operating lawn mower could be anybody. No, they need to have some training to know how they turn it on, or they turn it off, how they adjust it.

So we do have different level of modelers. While some modelers are going to be working on building their lawn mower, some other modelers will be using the final model or some applications.

We want to expand the usage of the modeling and this idea of model master file that people can go and reuse repeatedly will help us to achieve the goal. Otherwise, we will not be able to expand the usage of the modeling beyond the number of few modelers that we have.

And of course the models are becoming more complex, such as disease models, which are not just one or two equations. The disease model is different from the disease progression. Disease progression is just the one line. It could be even in straight line going down, but disease model is much more complex and requires lots of data unless you have got model master file that new group of experts sat down together and agree that would be better widely known.

So it's not just the owners of these model master file will benefit. Many, many people will be winner well. Whether they are commercial or non-commercial, there are patients will be benefiting, and regulators will be benefiting because it is much faster process of assessment. They will not be assessing every element of the model, they will be only focusing parts that somebody has actually adjusted.

So now you can see that on the back of what we talked about the open source and closed systems. Now you can see that not everything needs to be open source for certain purposes and how we are going to get that agree on it. And the reasons for that is clear. There are now evidence, particularly with these systems biology models, that 50% of the published models cannot be actually rerun. So therefore, in addition to data not coming from me, that's coming from others, but it's showing that we cannot actually rerun them. Forget about whether the results are correct or not. This is just ability to rerun and therefore we need some regulations around it. if we don't do this, modeling and simulation, in my view, will remain small, If we really want to expand the uses of modeling, we have to, get over this idea that everything needs to be open, etc. Because for some practical reasons we need model master file and some package that can be reused and reused without investigating every element. And this doesn't stop any academic research, because for academic research I told you anybody can use the open source versions of it and show the advantages and those advantages at some stage they roll into model master file.

Dr. Yuyan Jin:   I have one last question that is relevant to your presentation at the coming ISQP this year. We know that liquid biopsies enable doctor to discover a range of information about the tumors through a very small, simple blood samples. Can you share your view in terms of its application in addressing individual variability?

Professor Amin Rostami-Hodjegan:  Liquid biopsies are not more than patient characterization and patient characterization is something that we all have been working on using POP PK and all the other elements. In fact I would like to show a slide from a former Roche colleague, Bob power. The top corner that you are seeing is the typical kind of drug development that we will have. The blue area that you are seeing in that funnel are the patient groups that they get involved during the drug development until the drug comes to the market. The wider funnel are the patients who are really going to get the drug. FDA and others have recognized this with the guidance that came out in November 2020 titled “Enhancing the diversity of clinical trial populations-eligibility criteria, enrollment practices, and trial designs”, which encourages people to increase the diversity of patient populations. However, we know that has limits with regard to what we can include in any program of drug development.

From the end part of that funnel in this slide, you can see that the patients’ populations are very different from each other, and therefore it is challenging to say some of those patients will be actually exposed to the right level of the doses with precision and confidence.

Liquid biopsy is a technique that was used as a biomarker in the form of diagnosis for cancer. From there you are harvesting what is happening. A few years ago, we started to look into this to see if we can actually harvest something as a mirror of a tissue in the body where there is a tumor or whatever. Can we connect what we are measuring in this little fish that we have got from the plasma and linked to all the pool of fish that in that particular organ. Therefore, we started to match plasma and liver samples and measure proteomics in that issue and in the exosomes, we have separated using the techniques.

On the left hand of the side are the ones that we are currently using, like genotyping, phenotyping, and endogenous/exogenous markers. These are the bits that we are using in ADME to distinguish people from each other. Who is higher, who is lower etc. But the liquid biopsy is now giving the possibility of knowing exactly the level of proteins, which genotyping doesn't give you. Phenotyping only measure one thing at a given time. With liquid biopsy, we can actually measure array of different enzymes, transporters in one go and then characterize that patient. If you had got the model master file that we were talking about before, you can put all these into that model mast file and define what is the ADME for that specific patient. We really don't need to know or even think about their ethnicity, their gender etc. Because all of that is going to reflect on what we are measuring now. We have expanded this also to PD markers because we can expand our measurements to receptors and kinases etc. Therefore we don't need to have people as a yes or no for disease status. We can actually rank their disease based on abundance of certain targets that are changing and levels of kinases existing etc. I hope that I will be able to talk about all of these a little bit more during my talk is ISQP.

访谈译文：

金玉燕博士：您在FDA研讨会上展示了开放源代码，并接受了额外的访谈，以进一步开放论点。您能简要地与我们分享一下您对这个话题的看法吗？

Professor Amin Rostami-Hodjegan:  我想如果我可以分享一下我的屏幕会很有帮助。这是你提到的、我在FDA研讨会上展示的演示文稿。

出版社会支持开放源代码，主要有以下两个原因。人们对封闭系统的第一个反对意见是它是一个黑盒。这意味着我们对它没有信任，也是我们想打开他们并理解他们的原因。每个人都应该能够看到它。所以我们必须获取信任。另一个反对点是，如果有人想添加和修改这个代码，如果没有开放源代码，他们就无法这样做。

然而，我在CPT中发表的评论中提到：PSP是从一个单一角度出发的，当它们往一个单一方向发展时，必然有一些东西缺失。第一种情况是，当你试图向监管者提交一些东西或者你想做一个重大决定时，你必须相信建模和模拟的结果是可重复的。这种相信需要强烈到你可以相信，没有人可以在无人知晓的情况下影响到模型的任何元素。从这个角度来看，开放源代码有些脆弱，因为任何人都可以去更改它的某些元素。因此，我试图在这篇评论中解释的是，我们可以放弃现有的黑盒方法，因为其中的元素不可知，并转向一个我称之为玻璃盒的方法。玻璃盒意味着元素是已知的，但只有某些人可以改变它。

从一个学者的角度出发，我也赞成开放源代码。我的学生想尝试所有不同的元素。我们也特别需要通过开放源代码进行学术研究。但这里的底线基本上是在我们所说的“开放”源代码中找到一个平衡：每个人都可以访问一切信息的开放形式，和不是所有人都可以触摸它的封闭形式之间的平衡。某些特定的人可以看到部分信息，当然监管者或者需要评估它的人应该可以访问并理解它，但同时，玻璃盒子系统也可以摆脱开放系统的弱点。

金玉燕博士：我还有另一个相关的问题：你能和我们分享一下你对一个新术语“模型主文件”的看法吗？

Professor Amin Rostami-Hodjegan:  这是最近FDA研讨会上再次讨论的标题。Liang Zhao博士在我之前做了报告，反映了FDA的一些思考。我希望该专题介绍能很快完成。模型主文件包含的信息集并不是每个人都知道的，但现在我把它放在阻止我们从头开始的形式上，因为这是我们进入复杂的模型时遇到的一主要问题。

如果你从你的邻居那里借了一台割草机。与其给你割草机，想象一下他给了你割草机的不同零部分，说你自己做割草机。这是不必要的，很明显，他们可以帮助你节约时间，不仅给你零部件，也给你说明书。然而，想象一下说明书实际上并不与其匹配。理想的情况是直接把草坪修剪了。这并不意味着操作割草机的人可以是任何人。不，他们需要接受一些训练来了解他们是如何打开，或者如何关闭，如何调整的。

所以我们有不同级别的建模者。虽然一些建模者将致力于构建他们的割草机，但其他一些建模者将使用最终模型或一些应用程序。

我们希望扩展建模的使用，这种人们可以反复去重用的模型主文件的想法将帮助我们实现目标。否则，我们将无法将建模的使用扩展到我们拥有的少数建模者的数量之外。

当然，这些模型也变得越来越复杂，比如疾病模型，它们不仅仅是一个或两个方程。疾病模型不同于疾病进展。疾病进展只是一条线。它甚至可能是直线下降，但疾病模型要复杂得多，需要大量的数据，除非你有模型主文件，而这是新的专家组坐在一起，同意这将是更广为人知模型。

所以不只是这些模型主文件的所有者会受益。很多很多人将成为赢家。无论它们是商业的还是非商业的，都有患者会获益，监管者会获益，因为这个评估过程会快的多。他们不会评估模型的每个元素，他们只会关注有人实际调整过的部分。

所以现在你可以从我们讨论的开源和封闭系统的背面看到。现在你可以看到，出于某些目的并不是所有的东西都需要开源的，我们将如何就此事达成一致。而且其原因也很清楚。现在有证据表明，特别是对于这些系统生物学模型，50%的已发表的模型实际上不能重新运行。因此，除了数据不是来自我之外，这是来自其他人的，但这表明我们实际上不能重新运行它们。忘记这些结果是否正确。这只是重新运行的能力，因此我们需要一些相关的法规。如果我们不这样做，在我看来，建模和模拟将保持很小的部分，如果我们真的想扩大建模的用途，我们必须，克服这个一切都需要开放的想法，等等。因为出于一些实际原因，我们需要模型主文件和一些包，这些包可以在不研究每个元素的情况下重复使用。这并没有阻止任何学术研究，因为对于学术研究，我告诉你们，任何人都可以使用它的开源版本，并显示出优势，在某个阶段将那些优势整合到模型主文件中。

金玉燕博士：我有最后一个问题与您在今年即将举行的ISQP上的报告有关。我们知道，液体活检使医生能够通过非常小的、简单的血液样本发现一系列关于肿瘤的信息。您能分享一下您对其在解决个体差异应用方面的看法吗？

Professor Amin Rostami-Hodjegan:  液体活检不能逾越患者表征，患者表征是我们所有人一直在进行的工作，使用POP PK或其他方法。事实上，我想展示前罗氏员工Bob power的幻灯片。在幻灯片的上部分你们看到的是典型的药物研发历程。在漏斗的蓝色区域你们看到的是患者群体从药物开发过程直到药物上市所参与的各个环节。宽的漏斗是真正获取药物的患者。FDA和其他人已经认识到这点，并在2020年11月发布指南，标题为“增强临床试验人群-合格标准、招募实践和试验设计的多样性”，以鼓励增加患者人群的多样性。然而，我们知道在任何药物开发项目中可以纳入的内容都有局限性。

从本幻灯片中漏斗的末尾部分，您可以看到患者人群之间存在很大差异，因此说其中一些患者实际上将精确地和可信地暴露于正确水平的剂量是具有挑战性的。

液体活检是一种以生物标志物为形式的癌症诊断技术。从中你可以得到正在发生什么。几年前，我们开始研究这个，看看我们是否真的可以获得一些东西可以反映身体中是否有肿瘤或其他什么组织。我们能不能把我们从血浆中得到的信息与特定器官中的信息联系起来。因此，我们开始匹配血浆和肝脏样本，并测量这个组织和外泌体中的蛋白质组学，我们单独使用了这些技术。

幻灯片的左边是我们目前使用的，如基因分型、表型和内源性/外源性标记物。这些是我们在ADME中用来区分人与人的位点。谁更高，谁更低等。但是液体活检现在提供了准确了解蛋白质水平的可能性，这是基因分型不能给你的信息。表型分析只测量给定时间的一件事。通过液体活检，我们实际上可以一次性测量不同酶、转运蛋白的阵列，然后表征该患者。如果您获得了我们之前提到的模型主文件，您可以将所有这些信息放入模型主文件中，并定义特定患者的ADME。我们不需要知道甚至考虑他们的种族、性别等。因为所有这些都将在我们现在正在测量的东西中反映出来。我们也已经将其扩展到PD标志物，因为我们可以将测量扩展到受体和激酶等。因此，我们不需要知道其是否在疾病状态。我们实际上可以根据正在变化的某些靶标的丰度和现存激酶水平等对其疾病进行排序。我希望在我的演讲中更多谈论这些的地方是ISQP。

大会报告信息：

大会报告主题：The Holy Grail of addressing patient variability beyond University of genetics: liquid biopsy for patient characterization & as input to PBPK/QSP

（解决遗传学以外的患者变异性的圣杯：患者特征液体活检技术及与PBPK/QSP 模型的结合）

报告时间：12月5日下午，分会场二（定量药理学创新与特定疾病分会场）

会场主持人：庄笑梅 教授、李新刚 副教授

庄笑梅 博士

庄笑梅，博士，军事科学院军事医学研究院毒物药物研究所研究员，博士生导师。研究方向涵盖化合物ADME评价、新药临床前药代动力学研究、毒物代谢研究、药物相互作用及机制研究、以及PBPK建模和模拟。兼任北京药理学会药物代谢专业委员会主任委员、中国药理学会定量药理专委会委员、中国药物代谢专业委员会委员等职。主持国家科技重大专项以及军委后勤重大课题。以第一/通讯作者在Biochemical Pharmacology, Drug Metab Dispos, Toxicol Sci等杂志发表学术论文40余篇。获得授权国家发明专利6项。

李新刚 博士

李新刚，药剂学博士，副教授、副主任药师，首都医科大学附属北京友谊医院西药剂科副主任。研究兴趣：定量药理学与遗传药理学。他以第一/通讯作者身份发表SCI文章40余篇，总影响因子大于100分。

        2021年9月30日                优惠注册截止日期
        2021年12月5日-6日         大会正式议程（含现场注册）
        2021年12月7日                东亚论坛（线上会议）
        2021年12月25日-12月4日   会前和会后培训班
大会网站： https://isqp2021.sciconf.cn