[MUSIC] Today I'm going to start on a series of modules that discuss the genetic and other ways of personalizing therapy in a variety of therapeutic areas. So what I want to talk about is the way in which we understand the molecular genetic basis of disease, how those diseases respond to therapy, how we make diagnoses of those diseases using modern molecular tools, and how the treatments that we use can either be effective or cause side effects, and how we're understanding the molecular and genetic predictors of those kinds of outcomes. So the first area that I want to talk about is neuropsychiatric disease. Neuropsychiatric disease extracts a huge toll on people worldwide. These are two maps that show the impact of schizophrenia and unipolar depression measured as disability. So not prevalence, not how many people have the disease, and not how many people get the disease per year incidents, but actually what the impact of the disease is on those people. So Unipolar depression, a huge problem in the United States, huge problem in India and Western Europe. Schizophrenia, a problem worldwide including South America, Africa, and across Asia. So these are problems that are there. There's no question that there's a familial risk factor. So if you have a family member with schizophrenia or with depression you're at increased risk. So we think that there's a genetic story behind that. What is important to know is that it really affects people worldwide. Schizophrenia, around 0.5 % of people at some point in their life will be affected by Schizophrenia. Depression varies from country to country. 3% in Japan, up to 17% in the United States. An average of about 10% of patients at any given time have depression, so prevalence is about 10%. And in North America, in Canada, or the Unites States the probability of having a major depressive episode during any particular year is about 4% for men and about 8 to 10% for women. So these are enormous problems, and one of the issues is that the therapies that we have are effective in individual subjects, but there are many therapies that are ineffective in many subjects, so you don't know when you start treatment which treatment is going to be best for which patient. So, one question is, given the familial nature of these diseases, can we identify genes that are actually causative, or at least increase susceptibility. This is a genome-wide association study you're used to looking at these kinds of Manhattan plots now. This is a pretty dirty Manhattan plot and when you look at the Y access, none of those p-values are very, very, very low. There are many, many low psi in a huge number of cases suggest is that there are going to be genetic risk factors. They're going to carry very, very, modest odds ratios. Odds ratios are 1.1, 1.2 each, and the way in which they interact and which they drive the development of the disease Is something that GWAS, as I've said before and I will say again, GWAS is a first step to understanding fundamental disease mechanisms. So that's what this does, but it doesn't tell us anything about predicting disease in an individual subject. What is interesting, and what is really an emerging theme in neuropsychiatry, especially is a problem called copy number variation. Now the slide on the, the panel on the left shows a scan across a particular region of the chromosome, and each dot and there are many, many, many, many dots, illustrates the abundance of a particular gene or genetic region using snips, in fact, as the caller. So, in fact, when you look at the parents, the bottom two panels for these two particular regions that are being scanned they're all blue lines, and that means the parents have two copies of all the DNA that's being interrogated, when you look at the child, the child have blue lines all the way, but there's two places where there's less DNA. And the amount of DNA is decreased by 50%. So what that says is that in those stretches of DNA there's a big chunk, not a snip, not a couple of nucleotides, but hundreds, perhaps thousands of nucleotides that have just been deleted. And you can see because you understand the parents and child on this slide. You can see that these are both de novo changes, that is they occur in the child, but they do not occur in the parents. So if you look for de novo copy number variation, this is called copy number number variation, you can find a markedly increased incidence of copy number variation amongst schizophrenics, but what's really interesting is that you find a very, very high Incidence of copy number variation in autism. That's the red bars on the right. And so there's a lot of excitement about the idea that copy number variants, things that are difficult to assess using conventional technologies, may be a contributor to these diseases, and that would be an important step in terms of understanding their fundamental mechanism. Of course, if you have a copy number variation that stretches over thousands of base pairs, you are disrupting the function of many, many genes. So the way in which copy number variant produced the phenotypes that we're interested in, and those phenotypes appear to be largely, neuropsychiatric genotypes to-date, although there's not been a focus on copy variation in other diseases. It remains to be determined. So this is a big clue to a different pathway to understanding disease. At the other end of the spectrum is the idea of using next generation sequencing to understand rare diseases. This is a famous case that was reported a couple of years ago in science. There is a pair of fraternal twins that are indicated in black, so you can tell they're twins because they have that funny symbol on the family tree that they come from the same place, and they're a boy and a girl, so they have to be fraternal twins. They're both affected, and you can see there's the mutations that they have are actually listed underneath them. And they are, it turns out compound heterozygotes for rare variants in a particular gene that I'm going to tell you about now. They inherit one copy from the mother who's phenotypically normal, one copy from the father. And they have a disease which really resulted in severe movement disorders and severe intellectual disabilities. And so those are the kinds of problems that are really, really difficult to deal with. The fact that they both had it suggested that there might be a genetic basis. So using the kinds of approaches in next generation sequencing that I've talked about before, you sequence the effected children and you actually sequence the parents. And you come up with millions and millions of variants across the whole genome, but using filtering such as how many of them are encoding regions, who many of them are rare, how many of them have been seen before, how many of them are in places where you think you a mutation might cause a disease. You come up with a very, very small number of candidate genes, and actually it turns out that there's one candidate gene in these two individuals. Because what you're looking for here is two rare variants, one on one allele and one on the other allele. So, in these two children there was only one, and it was discovered in a gene called SPR. SPR is a critical step In the genesis, in the generation of Tetrahydrobioprotein is shown on this slide, and so when you have that in ways that we don't completely understand results in phenotype these children had. What was really interesting is that you can actually reverse the disease as you can see on this path way on the left lower side by administering L dopa. So these children, based on this genetic diagnosis, were given L dopa treatment, and have had dramatic intellectual and functional recoveries. So this is, going from a population, going from very large segments of genome deleted in diseases like autism, to a very specific set of genetic variants in a rare condition that offers a direct pathway to a treatment. So we have the spectrum in neuropsychiatry just like we do in other domains of medicine. I've shown you this slide before, just highlighting the idea that CYP2D6 genotype has a major affect on the doses that are required to treat with various antidepressants, depending on whether you're a poor metabolizer, an intermediate metabolizer, extensive, or ultra rapid metabolizer. And I've shown you this slide, the same idea for CYP2C19. This is another example, this is, one of the major effectors of neuropsychiatric function is serotonin in the cleft in between two neurons, and what controls serotonin concentration there largely Is a molecule called the serotonin transporter. The serotonin transporter gene is, the cartoon of its structure is shown on the top of the slide. And what's interesting is in the promoter region, the region of the gene that doesn't encode the protein, but actually controls how much of that protein you make, or is one of the major controllers of how much of that protein you make, has a common polymorphism is in it. The common polymorphism is called short or long. Short is a shorter version and makes less messenger RNA, less protein. The long makes more messenger RNA, and more protein. So people have asked the question, well if you have more serotonin transporter, does that mean you respond better to selective serotonin reuptake inhibitors, SSRIs, which are drugs like fluoxetine or paroxetine, very commonly used antidepressants? And the answer is maybe. So, all the dots to the right of this slide suggest in a trial, that the effect is better in people with, it's labeled here L/L, but it means, I'm sorry it's labeled II on the graph, but it's actually the long, long version. So those patients tend to respond better. So the way these graphs are displayed is that the mean affect in particular trial is the dot, the size of the trial is the size of the dot, and where their confidence intervals are shown, and where the confidence intervals intersect the vertical line, that means there's no statistically detectable effect. And so you can look at this kind of slide and you can see the results in ten different studies, mostly go to the right but on average there's not a big, big effect. So this an interesting observation, and whether you come away from this with the idea that this variant in some way affects SSRI response or not, depends on whether you're a genetic optimistic or not. But I think if, at best you can say that there's a very, very small signal. And of course there are some small studies that go in the opposite direction. There's some medium sized studies that go in no direction. On the other hand, there's a recent study that looked at response to Lithium, commonly used drug in bipolar disorder. And you're used to looking at genome-wide association studies now. This is a spectacular result with a p-value less than 10 to the minus 32. In a very small cohort of patients, 294 patients, treated with Lithium and then assessed on the basis of responders or nonresponders. The odds ratio for this signal is 112, and when they looked at the underlying genes at the particular locus the GADL1 gene was the candidate and resequencing identified a relatively uncommon snip that results in altered splicing of that particular gene. So this signal may all be related to GADL1 in ways that people don't completely understand. What's really quite interesting is reading the letters after this particular paper came out, and there is a number of groups around the world who have tried to replicate this and have been unable to. So you have a GWAS that shows a huge, highly statistically significant result, and yet others cannot replicate it. Of course, if others replicate it, then this would be an incredibly interesting important marker for driving selection of therapies in this disorder. If they don't replicate it, then we have to argue about why this signal came up in the first place. One possibility is the phenotype, what is it that you, how do you define response to lithium? Some people define it one way, some people define it another way. And it depends a little bit on the definition of the response phenotype. So, this is a signal that is very exciting, very new, and remains to be seen whether it can be implemented in clinical medicine or not. Clozepine is a widely used antipsychotic. It's highly effective thought to be one of the most effective antipsychotics we have. And the problem is, that it causes agranulocytosis which is a really, really feared side effect and can be fatal. It's never the drug of first choice because of that, and its use requires pretty frequent monitoring to detect agranulocytosis should it start to develop. What's interesting is that there's a pretty important, pretty impressive signal in the HLA locasis. It's not like some of the other signals I've shown you. It's not, the odds ration isn't 100. It's 17. But yet it does suggest that we could eventually develop a mechanism to identify subjects at particularly high risk for agranulocytosis and not use the drug under any circumstances. And then the rest of the population would still probably have to be surveyed, but you'd do that with the understanding that you probably have reduced the risk of this adverse event. Among antipsychotics right now, there's so called second generation anti-psychotic, which are highly effective and less toxic than initial drugs that were developed in the 1960s and 70s. Here is a set of drugs, and one of the side effects of these agents is weight gain. And it's particularly marked with a drug called Olanzapine where there's actually a reported increased incidence of diabetes. So, and each one of these drugs has other kinds of side effects so olanzapine still is one of the most widely used drugs, relatively well-tolerated except for this side effect, and what's interesting is that not everybody gains weight, but on average, people gain almost ten kilos. That's almost an enormous amount of weight, especially in patients with schizophrenia because the problem with schizophrenia, one problem with schizophrenia is that patients with that disease, which is a really devastating disease, tend to have a much higher incidence of, a much higher prevalence of risk factors for cardiovascular disease. They're sedentary, they're overweight to start with. And after olanzapine, they're more overweight. There is an increased incidence of hypertension, diabetes, and there's a very high incidence of smoking. So this is an undesirable side effect for many, many reasons. And what's interesting, to me at least, is when you start to do a GWAS on weight gain during olanzapine treatment in children, you get a very nice signal, and that signal is very close to a gene called MC4R. MC4R, rare variants in MC4R, are now known to be a cause of obesity, particularly in children. So you have this idea that there may be variation in this gene that doesn't necessarily cause obesity, but predisposes patients to much more weight gain during a stress like the administration of a drug that causes weight gain like Olanzapine. So whether, in fact, we can go from this kind of result, to screening patients at risk for massive weight gain during this drug remains to be seen. But this is an interesting signal in the right direction. So I've shown you this slide before, and I thought I would come back to this idea again, because over the last ten minutes or so I've shown you examples of very, very large signals that are associated with clear phenotypes, a rare variant in SPR that causes a dopa-responsive disease. That's very, very rare. Copy number variants that appear to be drivers of autism, and a variation that results in variability in lithium outcomes. On the other hand, the SIP2C19 and SIP2D6 stories that I've shown you before, particularly with drugs like, anti-depressant drugs like Venlafaxine shown here, the signals are much, much more modest. And so you have to ask yourself as we implement genomic medicine, are we going to focus on rare patients who have really extreme phenotypes and identifiable genetic variants that will be rare that actually could be actionable, or are we going to focus on large populations of patients with relatively common variation, and how we're going to manage delivering this kind of information to the healthcare system is going to be another challenge? So in summary, we have in the neuropsychiatric area, we have clear genetic susceptibility to disease and in drug response. Some markers have modest affect size. Others are very large, and may be clinically useful. And as I've said before, and as I will say again, understanding underlying pathways discovered through genomic and other mechanisms continue to provide insights into underlying disease mechanisms, and hopefully to new and targeted and effective drug therapies. [SOUND] [APPLAUSE]
