BROADCASTS.com

Published: Sept. 2, 2016, 8:51 p.m.

[intro music]

\xa0

Host \u2013 Dan Keller

Hello, and welcome to Episode Ninety-eight of Multiple Sclerosis Discovery, the podcast of the MS Discovery Forum. I\u2019m Dan Keller.

\xa0

Today's interview again features Dr. David Baker, Professor of Neuroimmunology at Queen Mary University of London in the U.K. We spoke at the ECTRIMS conference last fall. In part one of our interview he raised the issue of why there has been very poor translation from animal models to clinical trials. Today, Dr. Baker, also known as the \u201dMouse Doctor\u201d for his work with animal models, lays out why this situation exists and what to do about it.

\xa0

Interviewee \u2013 David Baker

I think there\u2019s many reasons why, and I think we all have our failings. And one can point the finger at the animal models, which a lot of the clinicians do, saying it\u2019s the animal model\u2019s fault, which is possible. But I think also we have to look at humans and how humans use their animal models. And then how humans translate the data from the animal models into the clinic, because I think there\u2019s many failings along the line, and I think that\u2019s one of the reasons for the failing between the two.

\xa0

I think one of the failings is, in terms of the animal models, that when we do our animal models for these, we\u2019re looking for mechanisms not treatments. And so about 70% of studies give drug before disease is ever induced, which never happens in a human. You know, you go after you\u2019ve had one or two or more attacks before you\u2019re given drugs. We also use the drugs in a way that are never used in a human, so people will do what they call a prophylactic drug where they\u2019ll give it before the disease manifests itself. Or a therapeutic dose, which is probably when the animals are showing their symptoms. But in reality, a human would be getting steroids at that time point. They would never get a DMT. So you\u2019re not comparing, you know, apples with apples. You\u2019re comparing apples with pears, and I think that\u2019s one of the problems.

\xa0

And I think, you know, if you try and block an immune response from being generated, that\u2019s quite easy compared to stopping an immune response once it\u2019s been generated, because immunity\u2019s about giving life-long protection against infections. And so I think it\u2019s a different type of beast to target. So I think this is where the animal models could do it, because EAE is one of the few where you have this relapsing-remitting disease course. But it\u2019s very, very rare that people actually start to treat in between attacks to block further relapses. I think that\u2019s one of the problems.

\xa0

The other big problem is the dose; the dose relationship between animals and humans. There\u2019s a tendency we just keep giving more and more and more and more, and eventually the drugs will work. But you\u2019ve got this problem that animals are very liable to be stressed, and we call it the building site effect, so construction site effect. And if you have lots of loud noises, it scares animals. They get very stressed, and your EAE just disappears. And likewise, if you just give lots and lots of drug, that probably tastes nasty. They get stressed out as well. And I think many of the so-called wonder cures \u2013 cures of the week \u2013 are because we\u2019re just giving too much, which doesn\u2019t have a relationship to what the human dose is going to be.

\xa0

And then, likewise, I think we\u2019ve got too much of a publication bias for the need to generate positive data. And I think what we then have to do is we have to look at the quality of the data. And I think there has been a lot of failure to replicate data. I think some of that is because some studies lack quality control, and the way I look at that \u2013 and I could be wrong; obviously it\u2019s an opinion \u2013 but if you look at the way that EAE is scored (it\u2019s normally a scoring system 1 to 3 or 1 to 4) and then you have your drug, which may be, you know, takes your control down from 3 down to a 1. But then, every now again, you look at the studies where it goes either way, and your controls are at 1 and it goes up to 3, and I ask the question how do you get a score of 1? Because if you had four animals, they\u2019re all scoring 1. Or is it three animals score 0 and one score 4, and that will give you a score of 1. And I think if people were made to actually put the data about how many animals got disease, we\u2019d be able to interpret those line graphs. Because I feel that, in many cases, some of those graphs lack quality control.

\xa0

If you have a robust quality control system, your control group should be giving you roughly the same type of scores every time. But in individual papers you can see, in some groups you have a score of 1 in the control group. The next experiment it\u2019s a score of 3. To my mind I think if you look at that, then those are probably the experiments are much more likely not to replicate. So I think you have to be, obviously, skeptical, but I really would like people to actually probably give us the information about how many animals got disease \u2013 what is their mean score \u2013 in addition to those line graphs. Because without that, they\u2019re impossible to interpret.

\xa0

So that\u2019s, you know, kind of one problem of the animals. And then for the humans, you have the same problems. So they over-interpret the animal data. The people doing the clinical trials are very, very rarely the people who came up with the idea. So if there\u2019s a weird side effect that you may know about, you know, that\u2019s not translated to the person who\u2019s actually doing the study, because they don\u2019t talk to the basic scientists. Then they probably underpower the studies. They don\u2019t necessarily pick the right outcome measurements. So I think there\u2019s many failings in both sides of the equation, and it\u2019s not always the animal model. But I think unless we kind of up our game, I think it\u2019s going to be very difficult for the people who are working on animal models, because you know, there are treatments that come along for, you know, the immune part of multiple sclerosis.

\xa0

And if you\u2019re thinking about the ethical use of animals, it\u2019s much harder to make the ethical argument that you should be using disease models which are very severe for the animals to try and work out fundamental parts of biology. And, therefore, I think we\u2019ll find that you know the funding agencies start to say, well, why are we funding this work? So I think we need to have good quality work, because if we don\u2019t have good quality work, it allows that clinical view that animal work doesn\u2019t really deliver the treatments. And I think they can deliver the treatments, but we just have to use our animal studies wisely to ask questions rather than, you know, blindly saying this will work in multiple sclerosis because it works in EAE. That doesn\u2019t make sense to me.

\xa0

Interviewer \u2013 Dan Keller

Do you have any succinct tips for people who are either reviewing papers on animal studies or people who are reading those papers once they\u2019re published or even the general public reading a news story?

\xa0

Dr. Baker

Well my first tip would be probably \u2013 and this is okay as an opinion \u2013 but, you know, EAE data is nonparametric. It goes 1, 2, 3, 4; it\u2019s not a continuous scale, so first tip is don\u2019t use, you know, the t-test of parametric data on nonparametric data. And that does make a difference. There is a Nature paper published this year that was analyzed with a t-test. If you analyze it with a Mann-Whitney test, which you should have done, the data becomes nonsignificant. So rather than the take home message is, you know, this is a new wonder drug for multiple sclerosis, their answer should have been you have to go back and reproduce your EAE experiment because it didn\u2019t work. So I think that would be the first tip. And then the second tip, I would really like people to say, tell us how many animals get disease and on what level and when, so we can interpret the line graph.

\xa0

MSDF

This is something that you routinely see in oncology done right. They talk about percent of responders, and among responders, what was the shrinkage of the tumor? They don\u2019t average it out among all the people who dilute it out by not responding.

\xa0

Dr. Baker

Well I think one of the problems as well is we\u2019ve also got this publication bias. We\u2019ve got you know this urge to see positive data, and I think that skews the whole system.

\xa0

MSDF

Has anything changed since you came out with a response to the animal checklist?

\xa0

Dr. Baker

I think, sadly, no, but we\u2019re actually doing the checklist again, so we will be able to see if things have changed. I don\u2019t think it has. I think the message hasn\u2019t gotten through. But I think \u2013 this is, again, another one of those nails in the animal model coffin that, if we don\u2019t up our game, we\u2019ll be seen to be doing an inferior quality work and eventually we\u2019ll get discarded. So I know that some of the grant councils are, as you know, saying this is a condition of your grant. But I think you know it\u2019s been slow to change, and I think one of the reasons is actually people who are leaders of the field actually are some of the people who are some of the worst offenders. So we\u2019re leading by bad example rather than good example.

\xa0

MSDF

We don\u2019t want to leave the listener with the impression that you\u2019re against animal models. I mean, you\u2019re known as \u201cDr. Mouse,\u201d so you know I guess you just want to see them done well.

\xa0

Dr. Baker

Yes, I\u2019m passionate. I mean, I really you know believe animal models have a real positive impact to do. And I\u2019ve been really lucky in the recent years is that, you know, some of those animal models \u2013 and work we\u2019ve done from animal models \u2013 is going through into humans and you know is starting to make the difference. So you know our work with the Cannabis was great. You know, it shows that you know our animal work has validity. Without the animal model stuff we\u2019ll never really understand the biology. You can\u2019t do all the experiments in humans. You do need experimental systems to be able to ask questions. And you need to be able to invent.

\xa0

And you know there is some fantastic work. You know I\u2019ve picked up the papers, and I get really excited by it, but I think, at the same time, we have to also be a drum to say, you know, try and improve the quality. Because, at the end of the day, it\u2019s more likely that if you\u2019re doing good quality animal experiments, that other people will be able to replicate it. And it will move the field on further and faster. And I think if people believe what we produce as being good solid work, then it\u2019s going to be a win-win situation.

\xa0

MSDF

It would be nice to see sort of a meta-analysis of animal studies that are considered to have been done well versus those not and see which ones translated into advances in the human situation, because so many times they say, well, sure it works in animals, but it doesn\u2019t work in humans. Well if it works in animals because it was set up not so well, then that might be a reason not to work in humans.

\xa0

Dr. Baker

Yes. I think you know the problem of animal models has got nothing peculiar to the multiple sclerosis field. It\u2019s just a common theme. And I think that tells me it\u2019s not a problem of animal models, because if it\u2019s so common in every other discipline, it tells us it\u2019s something how we use the animals is the fundamental problem. Now, you know for MS, we don\u2019t really know. I mean, I think this going to be the \u2013 we\u2019re at ECTRIMS now, and I think the whole world can change a little bit today, or in the next few days, because we\u2019ve always thought of MS as being a T cell-mediated disease. Now that may be still the true answer, but now we\u2019re starting to see ocrelizumab, which is a big B cell depleting antibody probably \u2013 I\u2019m predicting \u2013 to have as good an effect as anything that the T cell you know brigade has ever done. And, in fact, if you look at most of the MS drugs, you would say that most of them actually are inhibiting B cell function.

\xa0

Now, does that tell us that B cells are driving the disease? It may well do. Or it may well not. Now some people could argue \u2013 and they will \u2013 you know they\u2019re the reservoir for the virus that causes multiple sclerosis. And then other people will say, well, actually the antigen-presenting cells. And let\u2019s see, but I think what we\u2019ll find is you know EAE will have to have changed its focus. We\u2019ve been focusing our studies on T cell biology, but in fact, the T cell-inhibitory molecules haven\u2019t really delivered. So is that right? And it may well be you know we have to think of a different biology. But EAE can certainly do that if need be.

\xa0

So we\u2019ll have to you know try and work out how do these B cell-depleting agents work. Is it you know via antigen presentation or not? I don\u2019t know.

\xa0

MSDF

We\u2019ve always thought of T cells as regulating B cells. Now it looks like they both regulate each other.

\xa0

Dr. Baker

I mean, I have my history in skin diseases, and when I first started working, actually my boss was more interested in B-regulatory cells. T-regulatory cells kind of hadn\u2019t really existed at that time point. So I think we\u2019re trying to reinvent the wheel. If we look throughout the literature, it\u2019s a cross-talk between T and B cells are probably the answer. And we\u2019ll see. Again, from our animal studies, we\u2019ve had animal studies where we\u2019ve manipulated the immune system making sure that has a positive effect. We\u2019ve been able to translate that, so we have an N of 1 where we\u2019ve got rid of somebody\u2019s neutralizing beta-interferon antibodies by antigen-specific mechanisms. Now if we could translate that into MS, then we may have a way of treating MS. But we\u2019ll see.

\xa0

MSDF

Very good, thank you. I appreciate it.

\xa0

Dr. Baker

Okay.

\xa0

[transition music]

\xa0

MSDF

Thank you for listening to Episode Ninety-eight of Multiple Sclerosis Discovery. This episode is the final one in our series of MS podcasts. We hope that the series has been enlightening and has spurred further discussion about the causes of MS and related conditions, their pathological mechanisms, potential ways to intervene, and new research directions. We\u2019ve tried to communicate this information in a way that builds bridges among different disciplines, with a goal of opening new routes toward significant clinical advances. Although we won\u2019t be adding any new podcasts, the series will remain available on the MS Discovery website for the foreseeable future.

\xa0

This podcast was produced by the MS Discovery Forum, MSDF, the premier source of independent news and information on MS research. Msdiscovery.org is part of the nonprofit Accelerated Cure Project for Multiple Sclerosis. Robert McBurney is our President and CEO, and Hollie Schmidt is Vice President of Scientific Operations.

\xa0

[outro music]

\xa0

We\u2019re interested in your opinions. Please join the discussion on one of our online forums or send comments, criticisms, and suggestions to editor@msdiscovery.org.

\xa0

For Multiple Sclerosis Discovery, I'm Dan Keller.