Translating the 3Rs in complex physiological systems
It’s a great privilege to be giving this lecture. I’m going to use my studies, and those of colleagues, on nausea and vomiting to illustrate how we might apply some of the aspects of the three R’s to complex physiological systems. What I’m going to say also has some implications for studies of other complex systems, particularly those involving symptoms such as coughing, airway disease and a number of others.
I’m really very lucky to be talking at this exact time, because this is a convergence of a number of anniversaries. Firstly there is Darwin’s 200th anniversary since his birth coming up, and it’s almost 150 years since the publication of The Origin of Species.1 Next year is 50 years since the pivotal book – Principles of Humane Experimental Technique by Russell and Birch2, was published, which introduced the three R’s concepts that we’re now relatively familiar with. Lastly, it is 30 years since Professor David Smyth the physiologist published his book Alternatives to animal experiments in 19783. This book lists many of the alternatives that we would easily recognise today and Smyth is credited with introducing the term 'alternatives'. In Russell and Birch’s book, 'alternatives' does not appear in the index. I believe that physiologists are probably in one of the best places to implement and think about many of the aspects of the three R’s, particularly the aspects of refinement, as we’ll see.
The three R’s: here is the cover of the book (fig 1) and it’s no accident that there’s a table of random numbers in the background. They emphasise the importance of experimental design, which has now been embraced by a large part of the scientific community. Russell and Birch introduced the concept of the three R’s, replacement, refinement and reduction. The book links to Darwin because it includes quotations from On the Origin of Species at the head of each chapter. Russell and Birch said of Charles Darwin - “few people have been more concerned with the welfare of experimental animals or more active in furthering the progress of humane experimental techniques” - and I think Charles Darwin’s views on animal experimentation are a little neglected these days. I also think I’m correct in saying that Charles Darwin was the first honorary member of The Physiological Society.
Now, all this almost never happened, which gives me my link to nausea and vomiting.
On December 5th 1831 The Beagle was supposed to set sail but it was delayed because of bad weather. Darwin was very prone to sea sickness and wrote in a letter to William Fox (17th November 1831, letter 149) that “I long for the time when sea sickness will drown all such feelings and that time I do suppose will be the 5th of next month.” He constantly refers to being sea sick and there’s a rather nice description in one the biographies of him being “nailed to the rail” as the Beagle passed through the channel. It does make one think if he’d succumbed to sea sickness where would we be in terms of the theory of evolution; who would have substituted for him?
So I’m going to now focus, using those three books and some of the thoughts from them as the scaffolding, on nausea and vomiting as an example of a complex system to illustrate various approaches to the 3Rs. Some of what I’m going to say was encapsulated in a workshop supported by the NC3R’s that we had in the summer of 2007 (Holmes et al., 2009)4
So what do I mean by a complex physiological system? This is systems physiology or integrative physiology. They’re systems that involve homeostatic mechanisms and usually more than one organ or body system is involved - there has to be communication between multiple systems. Frequently the central nervous system is involved with outflows via the autonomic and somatic systems, but not necessarily both. Often the control is linked to a behavioural response such as drinking or escaping or vomiting and there may also be some sensory element to this, whether it’s fear, nausea or hunger. So these are complex systems within organisms. They also are often linked to defensive responses; fear, pain, sneezing and coughing, avoidance, learned aversions, nausea and vomiting all help the organism defend itself against various attacks. But many of these have a relationship to procedures which would be regulated under the Animals (Scientific Procedures) Act 1986 (ASPA), to humane end points or to symptoms and how they may be generated and treatments identified. I think there’s a very close link between looking at complex physiology and regulated procedures, and how we implement the three R’s when we’re investigating these physiological systems.
Vomiting is a very good example of a complex act because there really isn’t a single target tissue that you could look at in in-vitro studies, in contrast perhaps to the cardiovascular system. It involves a retrograde contraction in the intestine, relaxation of the stomach itself, contraction of the abdominal muscles and of the diaphragm to produce the final ejection. Control of this is incredibly complex. The pattern of motor activity in the abdominal muscles and the diaphragm is particularly complex and right at the last minute part of the diaphragm falls silent to allow the expulsion of material (vomit). That same effect occurs during belching and during gastro-oesophageal reflux. So there are some similarities here with other symptoms that may be of clinical interest. But the point to make is that there isn’t a single target tissue, you need the whole thing put together in order to be able to study the mechanism of vomiting.
All the motor activity is coordinated in the brain stem which sends the outflows to various parts of the body. The system itself is triggered by various inputs and of course any of these inputs could be responsible for triggering nausea and vomiting in response to an ingested toxin, to a drug as a side effect or to a symptom as part of a disease. Understanding these basic pathways and their inputs is crucial for understanding symptom generation or drug side effects. Of course there are multiple locations within the body, and also integrative centres in the brain, which are potential targets for anti-emetic drugs. So again one has to try for an integrated approach. Nausea is very poorly understood, but is known to require the rostral projection of information from the brain stem. We also know that in vomiting species a large rise in anti-diuretic hormone (vasopressin) is related to the symptoms of nausea. Nausea also produces aversive responses which may be an indirect way of measuring them. I think you can begin to see that this is a very complex piece of physiological integrative circuitry.
So how is this studied and why are we studying this? Well there are two primary reasons for studying this. Firstly, nausea and vomiting are common side effects of many drugs, including many that are already out there in use, and this affects patient compliance. It’s also a major impact on drug development; there are many examples, although not that many in the open literature, of agents which would be very useful for treating diseases but the side effect of nausea and vomiting is too great to be able to use them. We have to be able to understand what’s going on in that area in order to be able to remove it as a potential side effect - it’s a barrier to drug development.
We also want to understand nausea and vomiting because they are common symptoms requiring treatment. The clearest example I can give you here is the identification of anti-emetics to treat the nausea and vomiting which arise as a side effect of anti-cancer chemotherapy and also radiotherapy. There are two classes of drug that I’m going to talk about as we go, the substance P (NK1) receptor antagonists and the 5HT3 receptor antagonists. Both of these classes of anti-emetic are out there being used in patients and in both cases the work that led to their identification and development had a pivotal involvement of basic animal studies, particularly in the ferret.
Of course underpinning this is a basic understanding of the physiology and pharmacology of the emetic reflex. You need to know the basic science of this complex circuitry in order to understand how these agents cause the side effects and how to get rid of them, as well as to identify where in the circuitry one might be able to block in order to have an anti-emetic effect.
At first sight this looks like a very tall order for any of the 3R’s, but let’s begin to pick away at this.
Refinement: The choice of species
“Among the most important variables in the determination of procedures is that of the species of animal to be used.”
“It is just because it includes a wide knowledge of the special advantages of particular species for particular purposes, that a formal or informal training in zoology has again and again proved its metal in the progress of medical research.”
Russell and Burch 1959
Russell and Birch make a major point about the choice of species and I think we tend to forget this a little. We often use rodents but, as you’ll see in a minute, there are some issues with rodents and emesis which we ought to think about more broadly. They make a comment that zoology should be considered a little more, and while I’m not trying to introduce zoology back into first year classes, I do think there’s an element of people forgetting what species might be “available”. This isn’t new of course; Claude Bernard 5, in 1865 also made the point that “For each kind of investigation we should be careful to point out the proper choice of animal. This is so important that the solution to a physiological or pathological problem often depends solely on the appropriate choice of the animal for the experiment so as, to make the result clear and searching.”
In my area of research you’re limited in the species that you can use, the species that tend to have been used to study vomiting are dogs and cats, ferrets and more recently Suncus murinus, a type of shrew (house musk shrew)- these all have a vomiting reflex. The literature on dogs and cats goes back about 100 years, but increases around 1950. The ferret and the musk shrew were introduced in the 1980s. Pigs are now being used more commonly but there have not been a huge number of studies. There was always a steady trickle of non-human primate studies and more recently people have begun to look at another type of shrew, the least shrew, Cryptotis parva. So there are a range of species that have been used in emetic research but nevertheless probably the largest database is from dogs and cats. But the ferret has become more commonly used in more recent times, and I’ll return to the ferret shortly.
I frequently get asked, “Which is the best animal?” “Which is the one that’s most predictive of humans?” This is not an easy question to answer.
Even with something relatively simple, like apomorphine, that we know the pharmacology of and we know the site of action where it causes emesis, there is a range of variations in the sensitivity of species. For example the macaque is said to be not responsive, marmosets may be relatively weakly sensitive but dogs are very sensitive. The cat is relatively insensitive, there’s no obvious pattern but it illustrates the problem that you need to be very careful in selecting the species, so if people say “which is most predictive of humans?” You have to ask the question “predictive for what purpose?”
What about species lacking a vomiting reflex, could we use those instead perhaps of cats and dogs and ferrets? One of the things to ask about is how they differ, because if you’re substituting them and using a surrogate marker for vomiting or nausea, then you need to understand the general physiology of the individual species. Carnivores have an emetic reflex, clearly they can also belch, they have a gag and decrease gastric motility in response to toxins. Rodents (as represented by laboratory rats) don’t have a vomiting response and they are thought not to belch, which is consistent with what I said earlier about the diaphragm mechanism. So pica, measured via kaolin (clay) consumption, has been used as a surrogate marker for nausea or vomiting.
Now we’re not really sure if carnivores have pica in the same way as the rat. Rats certainly do, but the evidence that mice do is very thin and inconsistent. You can’t just treat a mouse as a small rat in this area. Another major difference between animals with and lacking the emetic reflex is that vasopressin ADH has a massive rise during nausea or prior to vomiting in species that can vomit. But in rodents there is not a large rise to similar stimuli, although oxytocin does go up. Oxytocin tends not to rise in carnivores, so this is a major species difference that we really need to understand. If you were using ADH as a surrogate marker of nausea, it may be fine in a carnivore, but it would not be appropriate to use in a rodent to see if there was distress from nausea.
The motilin system, involved in the regulation of gut motility, is present in carnivores but is incomplete in rodents. The ghrelin system is present in both and ghrelin itself can be anti-emetic. There are differences in the effect of 5HT on the vagus nerve; the percentage contribution to depolarisation from the receptor (5-HT3) involved in emesis is larger in species with an emetic reflex than in non-emetic species but this requires further study. There are also differences in the anatomy of the gastro-oesophageal junction. Again, I think people should begin to look at the anatomy more. Anatomy, rather like zoology, in this sense has been slightly lost.
Ferrets have played a major role in the study of influenza, the hypothalamic pituitary portal system, visual and auditory systems and also in emesis. I’m just going to focus on emesis, with examples. After chemotherapy and particularly cisplatin, which is often used as the prototypical example of an anti-cancer drug, vomiting occurs in two phases in humans; an acute phase lasting for the first 24 hours, then a quiescent period after which there is often a delayed phase going on for several days. In humans the acute phase is larger than the delayed, however, in the ferret it’s the other way round. But nevertheless, the ferret illustrates both phases, making it is an appropriate model to study at the mechanism of, and drugs which may work against, the emetic response to cisplatin.
A low dose of a substance P (NK1) antagonist has been shown to reduce the acute and delayed responses in ferrets, and higher doses completely abolish the effects. This NK1 receptor antagonist is currently in clinical use and improves the effects of the delayed phase in humans – demonstrating that results from tests in ferrets do translate to humans. Similar models were used to demonstrate the efficacy of the 5HT3 receptor antagonists that are also in clinical use. But NK1 receptor antagonists can block the emetic response completely in the ferret. So why doesn’t a complete block occur in all humans?
This issue of translation is a major one. One of the things that it probably tells us is that the involvement of the NK1 receptor and substance P in the brain stem in humans is not as great as it is in the ferret model. The ferret model predicted that the anti-emetic would work in humans but what it didn’t do was to tell you the exact detail of how well it would work. So whilst it’s a valid model, the precision may be an issue and we do need to better understand these translation issues.
It’s not due to binding with the receptor, because the ferret NK1 receptor is very close to the human NK1 receptor in the way that it recognises different compounds.
I believe that systematic reviews and meta-analysis are important in helping understand some of the translation issues in this area. My PhD student (Nathalie Percie Du Sert) has undertaken a systematic review as part of her project, looking at 5HT3 receptor antagonists and NK1 receptor antagonists in the ferret model over a three day period. And we can see from the analysis the NK1 receptor antagonist is an advance on the 5HT3 receptor antagonist, but this also enables us to get a better feel for how, at a detailed level, those efficacies translate into the clinical studies. We can get much closer now to comparing the animal data with that from human trials. It’s not easy to do, because you’re not comparing like for like, but remember that what we’re seeing here is a model of chemotherapy induced emesis, not a model of chemotherapy. The ferrets do not have tumours implanted, they do not have the concomitant medication that humans would do and many other factors differ. Recognising the limitations of the models is very important. However, the data is there for many areas to do systematic reviews and, subsequently, a meta-analysis to look at the efficacy. I think we should try and do this wherever the data is available.
So clearly in this very difficult system we’re looking at multiple inputs, integrative centres and outputs. Those agents I talked about, the NK1 receptor antagonists, work in the brain stem so it would appear that one would need the whole animal in order to show that those agents can produce a block of this reflex. But you’ll see in a minute that there are ways of applying refinement to those studies; nausea is difficult to measure in animal studies and this itself leads onto aspects of refinement.
Nathalie Percie Du Sert has been working with Professor John Rudd at the Chinese University of Hong Kong, looking at the use of telemetry in the ferret and measuring the electrical activity of the stomach – the gastric myoelectric activity (GMA) which when measured using cutaneous electrodes is called the electrogastrogram (EGG). We know what happens to this in humans under various circumstances including nausea – there is a shift from the dominant frequency and a disruption of the rhythm. A similar disruption of the rhythm occurs as a response to emetic stimuli in the ferret. Not only can you monitor the electrogastrogram which gives you something that then you can compare with studies in humans to look how good the model is, it may give you insights into sub-emetic end points so that you may not have to induce emesis, you may produce similar changes at sub-emetic doses and therefore have a more refined end point. But also if there is vomiting, because this technique can also measure pressure in the abdomen, you can use it to identify all the episodes of retching and distinguish them from vomiting and defecation by the pressure profile. It’s possible to automate the analysis of this, which although still under development gives an idea of how one can refine in this particular area and gather a much larger amount of data.
Professor Rudd (Hong Kong) and I are interested in the anti-emetic effects of a range of vanilloid receptor agonists. Blood pressure and core temperature recording by telemetry shows that one agonist produces a transient elevation of blood pressure and also a fall in temperature, whereas the others don’t. Using this type of analysis may help you to identify compounds which are producing undesirable effects, or perhaps which may confound your primary action, and to select compounds which do not have adverse effects such as elevation of blood pressure or a fall in temperature. I doubt that one would have picked this up by almost any other method because the change in blood pressure is so transient.
Russell and Birch only really refer to nausea or visceral illness at a couple of points in the book, but they certainly talk about visceral ill health being discomfort if not pain, and curiously they mention about nausea in pigeons. You’ll notice I’ve very carefully talked about measuring vomiting, not about measuring nausea.
"Nausea is a thoroughly distressing state in man, and by human analogy we might well suppose it to be so in pigeons."
Russell and Birch 1959
Nausea is an issue because, as this rather nice headline says, we don’t know what the animal is thinking. I do wonder whether nausea is an unrecognised symptom in experimental animals and something we should pay more attention to. There is an analogy here - I don’t think you as an observer could be any more certain whether a baby has nausea any more than you can tell whether a particular animal does. It’s a deduction. I think it’s worth pointing out that this isn’t just related to emesis mechanism research, because any of the drugs in development, for any kind of application, could have nausea as an underlying effect. How would you ever pick it up?
One way may be to measure pica by kaolin consumption. Cisplatin is a cytotoxic drug that markedly stimulates consumption of kaolin in a dose-related manner in rats. There is also a gradual reduction in food intake, which you would pick up anyway, and there is a delay in gastric emptying. If the kaolin was not there and you were giving the animal a novel drug as part of your protocol, you would not necessarily have any indication, other than reduction of food intake, that there was an issue.
The measurement of gastric contents is something that’s ideally suited for refinement because in the above studies the animals were killed, control and drug, in order to look at the change in the shape / size of the stomach and the degree of change of emptying. However, you could now do this by using imaging techniques, so even in the span of a few years techniques have moved on considerably. Delays in gastric emptying and consumption of kaolin combined with food intake may be surrogate markers in rodents. So the potential utility of rodent studies in this area should perhaps be reviewed.
I want to come back to Russell and Birch because they talk about replacement in conscious living vertebrates, but also because they use the phrase “comparative substitution”, which means using less sentient or invertebrate species. I think we should revive this term, rather than use “pure alternative”, when we’re talking about an invertebrate. Studies in invertebrates have been the origin of a number of discoveries which have eventually had clinical impact.
I want to illustrate to you an approach, which uses non-animal models, that has been developed to identify potentially emetic compounds early on. Dictyostelium, which is a “social” amoeba, has been used to study the way that some drugs work (e.g. valproic acid for epilepsy) and its one of the NIH approved validated models. Preliminary studies in collaboration with Dr Robin Williams and Ms Janina Mukanowa at Royal Holloway University of London, funded by UFAW, have shown that Dictyostelium respond to denatonium, a T2R receptor agonist. T2R receptors are the bitter taste receptors on your tongue. They are also on some of the cells in your GI tract that are involved in making you vomit in response to certain compounds. We’re looking at a whole range of compounds to see whether this organism could be used to detect those which have the potential to induce emesis or to have other related side effects. I should emphasise that these studies are not aimed at developing new anti-emetics.
If you want to look at complex physiological systems, you’ve got to have some complex physiology. I think the unicellular organisms and “lower” invertebrates, if I can use that term, are not particularly helpful. Organisms like the holothurians, which have a “proper” gastrointestinal tract and collections of nerve cells are more suitable. They also have a type of vascular system, so one can conceive of doing more classical physiology on these types of animals. Sea cucumbers for example have one useful property; they eviscerate their entire GI tract when attacked and then they can regenerate it, including the neurones. Understanding this might give a very interesting model for some regenerative processes.
Another species that one might look at would be the cephalopods. I think there are some very interesting questions, particularly related to learning, memory and the central nervous system that could be looked at in these creatures, but we’re going to need a whole other way of assessing humane end points. The organisation of their central nervous system is very different but it is lobular; they have lobules and functions residing in particular regions of their central nervous system. But two thirds of the neurones are in the arms, and the arms are capable of all sorts of interesting reflex responses. So how will we assess the classical indicators of pain such as withdrawal, when a lot of this may be relegated or devolved to the arm nervous system? And the optic lobes also contain a large proportion of the neurones and a very large amount of processing of visual information is done outside the central part of the brain, which actually has got the lowest number of neurones. But there are a whole range of surgical techniques that were developed in the 50s, 60s and 70s (e.g. decerebration, vascular catheterisation) which may need to be revived, revisited and refined if we are going to start to use these species as replacements for some of the “higher” animals. There are very interesting questions to be asked about cephalopods but it would need a whole different approach for looking at pain, distress and lasting suffering I think, in these creatures.
Russell and Birch use absolute and relative replacement and I think we need to readdress this. Relative replacement is where animals are still required but during the experiment they are probably or certainly exposed to no distress at all, this covers anaesthesia, CNS destruction, decerebration and other procedures and Schedule 1- although Schedule 1 procedures require some sort of target tissue or organ to be of real benefit. But although the studies using in vitro animal tissue obtained by Schedule 1 methods are very useful the replacement agenda should include the use of more human tissue obtained during surgery. I think there’s a lot that could be done on human tissue and it’s an underexploited resource, but again there are many ethical issues associated with that as well as with animal research.
Things have been done in my own area of emesis research using non-sentient animals. Professor Julian Paton (Bristol) and Dr Julia Smith (St George’s) developed a working heart decerebrate shrew preparation almost 10 years ago. This complicated preparation can be used to look at emesis, but it’s a time consuming, expert type of procedure. It’s suitable for the study of the physiology of the brain stem, but it won’t be used for identification of anti-emetics or looking at emetic mechanisms for systemically active agents.
Areas that would be very difficult to replace are those which model the effect of a surgical procedure such as Nissen fundoplication, where part of the stomach is wrapped round the lower oesophagus in order to reduce reflux. The ferret has been used as a model to look at this and this procedure was found to sensitise the emetic reflex. This has helped understanding of some of the things that occur as a side effect of this procedure in humans. So it’s very difficult to see how one could model these in any other way except in a whole animal setting.
Pathways have been studied to determine the mechanism by which chemotherapeutic agents cause emesis, and this is another area that is difficult to replace. This research involved nerve lesion studies, cutting the pathway - the vagal afferent neurones that convey information from the gut to the brain. In this pathway the enterochromaffin cells release 5- hydroxytryptamine from the gut, which fires the vagal afferent and activates the 5HT3 receptors located on the vagal afferents. The effect of these cells can be blocked by a 5HT3 receptor antagonist acting in either the gut or the central nervous system. Essentially, these studies show you the basic site, and the in vivo study can inform
in vitro studies – demonstrating which tissues you could look at peripherally to target. So for example one could envisage how an enterochromaffin cell culture looking at the release of 5HT could be utilized. If your compound induced a large release of 5HT it’s likely to have an emetic liability. So these whole-animal nerve lesion studies can be used to identify ways to develop replacement strategies and use tissue cultures to study potential targets in this area.
Absolute replacement is where animals are not required at any stage, which is difficult in this area of pathways for nausea and vomiting, and identification of anti-emetics. These are problematic areas for complete replacement, but identification of emetic liability of new compounds is an opportunity, and I think it’s an area that we’re gradually getting to grips with.
Human studies have a role here, and although they’re very difficult because of recruitment and other issues, I think they need to be considered. I am aware of only one study aimed at identification of the brain areas activated in a subject with nausea - we need far more of these. There are ways of studying emetic agents in humans and of course motion sickness is an option. They are difficult studies to do but it doesn’t mean we shouldn’t try to do them with appropriate regulation. We need to develop these models that can be more used in humans.
There is a need for a shared information database in developing 3Rs options in the identification of emetic liability in new chemicals, as this would enable access to all information at just one source. There is an enormous wealth of literature out there, both in the public domain and the pharmaceutical industry and I call these the “unknown knowns”; it’s there but people don’t know it’s there and it’s not gathered in one place to use. It’s easy to do this in emesis because it’s a very well defined area of research. It’s of high importance to drug development because it can hold back progress in some areas.
There’s human data, but in many cases the human data will never be repeated - it’s wasted if it’s not incorporated into a database. If we really understand why substances are emetic and where they are emetic, this will give us insights into novel anti-emetics. And you can see how this might all fit into a cascade to look at the emetic liability 4. You would start by interrogating a database and the basic pharmacological properties. You might then go to a micro organism like Dictyostelium and then to a human cell line for example an enterochromaffin cell to see what the liability is for 5HT release. All of these would generate a score incorporating a dose-response liability. Overall the score would identify, before going in vivo, the probability of problems developing. It will also give you greater confidence and may help you to reduce the number of animals that you use. Or perhaps you say - ok, we’ve got a high liability at this point, so we will just do a study in a mouse looking at gastric distension and reduction of food intake, and if that then gives you a positive (i.e. stasis and reduced food intake) you do not decide to go any further. Alternatively you might then use telemetry to look at sub-emetic biomarkers in relevant species - it’s a cascade of testing. This process is about utilising all the information we already have and undertaking in vitro studies (preferably using human tissue) prior to undertaking in vivo studies starting with the species of lowest sentience.
So, to the future: what will it take to make advances? I think researchers should view the three R’s as an opportunity, not as a threat, and engage. It is also an intellectual challenge for academics and universities. Learned societies have an important role to act as catalysts via meetings and education.
Darwin, C. (1859) The Origin of the Species. John Murray. London.
Russell, W.M.S. and Birch, R.L. (1959) The Principles of Humane Experimental Technique. Methuen &Co Ltd. London.
Smyth, D. H. (1978) Alternatives to animal experiments. Scolar Press in association with The Research Defence Society. London.
Holmes, A.M., Rudd, J.A., Tattersall, F.D., Aziz, Q. and Andrews, P.L.R. (2009). Opportunities for the replacement of animals in the study of nausea and vomiting. replacement in a multi-system reflex. Br J Pharmacol, 157, 865-880.
Bernard, C. (1865). An Introduction to the Study of Experimental Medicine. H.C. Green (trans., 1949). New York: Henry Schuman.