Email Icon Facebook icon Twitter Icon GooglePlus Icon Contact

User Top Menu

SRF Interviews Tom Insel

Posted on 8 Aug 2007

The current inhabitant of the NIMH hot seat is Tom Insel. Insel is in charge of a budget in the $1.5 billion range, and while he answers officially to the executive branch, he must also mediate the unofficial demands of legislators and numerous interest groups. He manages this with a thoughtful, unruffled air, never failing to be courteous to everyone from poobahs to administrative assistants. We're delighted that he took the time to sit down for a long telephone interview with SRF editor Hakon Heimer.

Note: As with other areas of the SRF website, you can submit comments on this interview. To the extent possible, Insel will reply to questions raised in these comments.

SRF: People know you as the NIMH director, but what is the path that took you to this position?

TI: It's completely nonlinear. I started in a highly psychoanalytic residency program and actually did a Jungian analysis when I was at the University of California, San Francisco as a psychiatry resident. Much of my training involved doing long-term treatment of a whole range of patients at UCSF, with very limited exposure to pharmacotherapy, which was not a major focus of training in that department in the 1970s.

I came to NIH mostly on impulse at the end of residency, and became for the first time really excited about the opportunity for innovative treatment development and for studying psychiatric illness not just from a psychodynamic perspective but from a biological one, and most of all interested in how to put those two together. I focused initially on obsessive-compulsive disorder, partly because nobody else was doing it, and also because there were patients available who were adults that couldn't fit into Judy Rapoport's study. Judy was the first person to really develop OCD as a research focus in the United States. And I got lucky: at that point we were just seeing the advent of drugs that might have more of an impact on the serotonin reuptake system and discovered that they seemed to be particularly useful for people with OCD. So my group chased that for about 4 years. We did a whole series of obvious studies: clinical trials, CSF differences, endocrine measures, sleep studies—all the things that were then available at the NIH. In 1983, it seemed that the future was going to be just following up on those initial descriptive trials. I wanted to do something that would get more at the mechanisms of illness.

At that point—less than 25 years ago—we didn't really have much available for doing studies of the brain. So I moved away from work with patients and took a sabbatical at Hopkins for 6 months in chemical neuroanatomy with Michael Kuhar. From there I decided to develop my own laboratory, trying to understand the neurochemical basis of complex behaviors. I decided that we'd start someplace where nobody else had been, so that happened to be in looking at the neurobiology of positive relationships: parental care, infant responses to brief social separation, and later on social attachment.

SRF: So you were working in humans at first, or you started this new direction in animals?

TI: No, I moved from humans because it's so difficult to do those sorts of mechanistic studies in humans. I really wanted to get at the molecular and cellular basis of behavior. To identify the brain regions and the major neurochemical systems involved in social behavior, we used parental care. Pedersen and Prange had demonstrated that oxytocin, one specific neuropeptide, seemed to be important for parental care in rats (Pedersen and Prange, 1979). So we tried to understand: how was it working? Where was oxytocin binding in the brain? What was it doing there? What were the pathways that were important for regulating oxytocin and its effect on the onset of maternal behavior in rats? And that turned out to be a really fascinating topic because by the late 1980s and early 1990s, we already had some very good tools for answering these questions.

I got to a certain point where I became more interested in how males and females become attached to each other. We couldn't really do those studies in rats or mice because they don't develop those kinds of selective, enduring attachments, so my lab moved into studying voles. We spent quite a few years trying to understand species differences: why some species are monogamous and form long-term attachments and others are solitary without selective social attachments. Once again, oxytocin and its sister peptide, vasopressin, turned out to be really interesting factors.

I moved to Emory in 1994. At that point the work really had to incorporate more molecular approaches. I was incredibly fortunate to have Larry Young join my lab as a postdoc. Larry brought powerful molecular skills to the same questions. We were able to do far more precise studies using transgenic mice: what happens if you knock out oxytocin? What happens if you overexpress vasopressin or its receptor? Over the course of 5 or 6 years we were able to put together one of the more interesting stories about the molecular and cellular basis for a complex behavior in mammals—the role of both oxytocin and vasopressin in mediating attachment behavior in the prairie voles.

In 2002, I got a call asking about my interest in the directorship of NIMH. I wasn't looking but I was interested. Frankly, I'd gotten to a point in my own life where I thought, well, it's really exciting to have a lab that's productive, to be making progress on an interesting scientific problem, and to have papers in Nature. But I began to feel like something was missing. I wondered, does this really make any difference to anybody? And I really wanted to recapture part of what drove me into medicine in the first place, which was trying to help very ill people. That's what really brought me back to NIH. Here we are driven by this public health mission, thinking about how to apply these very powerful tools for studying the brain so that we can understand schizophrenia, bipolar illness, mood and anxiety disorders as brain illnesses. Returning to psychiatry research after 20 years in behavioral neuroscience, I was surprised that progress wasn't greater. We still have only a primitive understanding of the pathophysiology of these disorders. And during my 20-year sabbatical, when breakthroughs were evident in other parts of medicine, where were the innovative, effective treatments for mental illnesses?

SRF: Since you stopped working on social attachment and these hormones, what have you seen that's been really interesting that's come out of that research field? I'm particularly interested because there have been some recent papers on social cognition and schizophrenia. I'm curious if there's work that's synthesizing this from the animal to human level, and then possibly into areas of schizophrenia psychopathology other than just positive symptoms.

TI: I think most of the work that's built on what we did in animals has focused more on autism. These highly conserved systems—which we know go all the way back to invertebrates—that are so important for social interest and social behavior, is it possible that they are somehow defective in kids with autism who don't have fundamental social interests and don't develop typical social behaviors? I think the jury's still out on that. There are a couple of fragments of information consistent with that hypothesis, but we don't really have the data we need, and certainly what would be most interesting would be developing novel therapeutics for social behavior based on this information. That we certainly don't have yet, though a few people are working on it.

For schizophrenia I don't think there's been much that has built on the work that we had done in the voles. Probably the more relevant piece is my second line of research, which has gotten much less attention but I think is probably much more relevant to schizophrenia, and that had to do with epigenetics. We were one of the first labs to begin looking at why we were getting such striking individual differences in behavior of our isogenic mice, depending on how they were being raised. So we began trying to understand how the environment alters patterns of gene expression and alters patterns of behavior (Francis et al., 2003). We took blastocysts at the eight-cell stage and we cross-fostered them. This is a prenatal cross-fostering approach, from one strain of mouse to another and we followed this with a more classical postnatal cross-fostering step as well. The resulting phenotype depended a lot on which uterus the mice grew up in and on postnatal experience, more than on their genotype. That finding, to me, keeps us honest because it reminds me that, as much as we've focused on genetic sequence for the associations for some of the mental illnesses, we've also got to be very cognizant that genes are only going to explain some fraction of the variance here. Early experience, and by that we may be talking prenatal experience, could be very important for not only patterns of gene expression through epigenetic modifications, but ultimately cognition and behavior.

SRF: Let's shift to NIMH and the job that you do now, starting with the big question on the minds of most U.S. researchers: is there anything positive to report on extramural funding since your last update of about December 2006 (see SRF related news story, as well as an SRF news story on a funding session at the 2006 Society for Neuroscience meeting)?

TI: If you look at the 2008 budgets that are being discussed, the most ambitious one that's coming from the Senate side is a 2.2 percent increase. The House looks like about a 1.2 percent increase. Both of them are more than the president's budget, which is about a 0.1 increase: a basically flat budget. The rate of inflation, the Biomedical Research and Development Price Index (BRDPI), last year was about 4.5; this coming year for 2008, the prediction is about 3.6 or somewhere in that range. So with any of the proposed budgets, we're still below inflation, and if in aggregate you look over the last four, we've been falling below inflation in every year. It now amounts to about a 12 percent loss of purchasing power from just the inflationary increases in the costs of doing science.

Having said that, there are a few glimmers that make this still workable for us. One is that, at least in 2007, the costs of the NIH Roadmap were paid for directly out of funds that were appropriated to the NIH Director rather than from the individual institute's budgets. We had budgeted about $18 million this year to pay for our Roadmap projects, so that was money that essentially came back to us to be able to spend in the 2007 budget year. It still keeps us a little below inflation, but it certainly mitigates what looked like it was going to be a dire situation.

The other thing that's helped us make it through this 4-year period is that we had very large contracts for research and clinical trials, like the CATIE trial, which ended during this period, and so some of that money was freed up and has helped soften the hard landing after the doubling of the NIH budget. So we've been able to support more new research projects during a time when some of our sister institutes have had difficulty. We've generally been able to pay about three-quarters of the grants under the twentieth percentile, and that's better than many of the institutes at NIH that are dealing with pay lines that are in the 10 percent range.

SRF: A related question having to do with who gets what, given a limited pie: a big news story in late 2006 was that there was a lot of pressure being applied by different interest groups to Congress to add $100 million for autism research to the NIH budget, if I remember the details correctly. The criticisms that I heard at conferences and in the media were that, first, the money might come out of budgets for other disease research, and, second, that it wasn't clear that even if the money were new money, that this would be the best place to put these amounts. Either autism research wasn't a mature enough field that that money would be justified, or simply that it should be up to the scientists to decide what was the best basic research that would benefit everybody. Where does that situation stand now?

TI: There was a bill called the Combating Autism Act that was passed by the House and Senate and signed by the president on December 19, 2006. The Act has a number of things, including some budgetary recommendations; however, those are what we call authorizations. They're not appropriations, and the difference is that you can authorize that $100 million should be spent on autism, but unless it's appropriated, there's really no money with which to do that. And what the language ultimately said in the Combating Autism Act is "funds as appropriated" up to the following level, which in this case is $100 million. So, it is now up to the appropriators in Congress to fulfill the suggested authorized amounts in the Combating Autism Act. It is important to note that, in fact, NIH is spending more than $100 million on autism this year, so this part of the Act should be fulfilled even without funds coming from other research commitments.

SRF: Fuller Torrey and others criticized the NIMH strongly a few years ago about funding priorities, that is, that serious illnesses with serious impacts were being underfunded in favor of things that were considered less important. Did you agree with that criticism and do you see that things have changed substantially, and/or is there room for improvement on that front?

TI: I agree with Fuller that there were some areas that we had been funding for many, many years, and it wasn't clear that continuing to pursue those questions was going to give us more than minimally incremental information, and there are other areas, particularly some of the areas around serious mental illness, where we could do more, and we could do better. A lot of this is really trying to find a balance—the sweet spot—between two factors: you want to find the area where there's the greatest public health need and where there is scientific traction for progress. There is to me no question that with genetics, imaging, some of the work that can be done at the cellular level, there are opportunities to really make headway on pathophysiology and new treatment strategies for these very complicated problems, opportunities that we did not have a few years ago.

SRF: As you probably know, a common topic at the lunch tables at schizophrenia conferences is the limited pot of money for schizophrenia research, and particularly schizophrenia genetics research. There is a complaint that it's being concentrated in too limited a scope, for example, whole-genome association or other "big" technological science, and a related complaint that it's being concentrated in too few hands. A few recent prominent examples are the large gifts from Stanley Foundation to Broad Institute and Cold Spring Harbor Labs for psychiatric research. I know that money follows money and people bet on winners and this is often a good strategy, but is it necessarily a good thing that we put a lot of faith in whole-genome analysis, or is there a danger of good science in small groups at some small universities losing out on the money, and if so, how can that be mitigated?

TI: In a way, this question is a subset of the bigger question of big science versus small science. We deal with this every day in almost every area that we work in, both clinical and basic. Part of the skill of running an institute like this, investing about $1.5 billion, is finding the right balance between big science and small science, because they do different things. Genetics is the perfect example of big science where you have, I think, right now an opportunity to quickly map all of the common points of variation, the common SNPs and the copy number variations, quickly in large numbers to find out whether there is a signal for schizophrenia or for bipolar illness. There are hundreds of thousands of variables, so to have decent statistical power you need thousands of patients and thousands of controls, you need to have groups that are willing to work together, you need to have someone to coordinate it all, and you need to be able to come up with significant resources to be able to get the whole-genome association studies done. Much of the genotyping costs for whole-genome association of NIMH samples is being supported by the Foundation for the NIH via the GAIN project, so it's not actually coming out of our budget. We're providing the samples through the NIMH repository—that piece we did pay for—to build a repository of tens of thousands of samples with all the phenotyping data, but it's a public-private consortium working with FDA and NIH through the Foundation of NIH that's actually doing the GAIN Project.

A really important thing for people who are part of the Forum to realize is that this is all going to become public, and in a sense it's like the Human Genome Project. We need to get through this stage quickly, and this is being done by a small number of groups using this repository and using other samples as well. It will all be public the day that the genotyping is cleaned, and there will be an opportunity then for everyone anywhere to begin to mine those data for their favorite sequence variations to see whether there is an association if you modify the phenotypes and the genotypes in different ways. So all of a sudden, it will be a place where large science will fuel small science—small labs will be able to work in this arena.

It is important to put all of this in context. The whole-genome association is really a screening effort; it's just a way of getting a first pass at where the lampposts are. Where are the places in the genome that we ought to be focusing on to look for rare mutations or polymorphisms? People will have to do a considerable amount of resequencing, and that's going to become much more accessible and much less expensive, with the advent of these new systems for high-throughput sequencing. But even more important than finding the gene associations is going to be taking this from the level of statistics to the level of biology. That's what really needs to happen, just as it has in the rest of medicine. If the schizophrenia community really wants to go for the brass ring, they will find out where the important areas of the genome are quickly, and then do the heavy lifting of finding out what those have to do with brain development and how you go from statistical genetics to neurobiology. That's actually going to be a much more intellectually invigorating and much more insightful process.

Technology will drive much of this progress. For instance, sequencing is going to be about 1 percent of what it was 5 years ago in terms of costs, and it'll be about 100 times faster. What I'm hoping people will start to do is to move beyond the sequencing stage and think about function. If you put an allele into flies, or you put it into zebrafish, or you put it into a mouse, how does it change brain development? Where is it expressed, and what's its function? And is this going to be a potential new site for us to be thinking about interventions? Could it ultimately be a biomarker so that we could identify who actually has the prodrome of this illness and maybe preempt the psychotic phase of the illness? I think that's where the excitement will be. For many of these sequence variations, we're talking about resulting splice variants, some of which are brain-specific, some of which may be even human-specific, and some of which are clearly heavily expressed in the parts of the brain that we're most interested in for schizophrenia. The neuregulin 1 type 4 isoform that was published last week in the Journal of Biological Chemistry [Editor's Note: see SRF related news story], this is, to me, an extraordinary insight, and I think it's a great model for where the field is going to have to go. Here you've got a SNP that has been associated with schizophrenia, and the SNP turns out to be a functional polymorphism, and it sits in the promoter for an isoform that is brain-specific, and it's heavily overexpressed in human fetal brain. So this begins to tell you something about where you want to be looking and how you might want to look at the function of this polymorphism that is keying up brain development in such a way that someone is at a higher risk of developing a psychotic illness.

SRF: I have heard pointed comments that the recent Wellcome Trust study of seven major diseases, including bipolar disease [Editor's Note: see SRF related news story], shows that the WGA approach is not as powerful as it's cracked up to be. Another, and perhaps related comment that I've heard several times is that the success of psychiatric genetics to date derives from smaller, more homogenous samples, as opposed to collapsing these 500 or 300 group samples into a big 2,000-person sample. Do you want to address either of those?

TI: They're related, in a way. I think the important thing here is to manage expectations about whole-genome association. I don't think anybody could seriously argue that this hasn't been revolutionary for macular degeneration, for prostate cancer, for irritable bowel syndrome; so we've got some striking success stories over the last 12 months. The remaining question is whether we'll see anything like that for any of the psychiatric disorders. We may; we may not. It could be that the common variations that you see on the Illumina or the Affymetrix platforms just aren't the story for these disorders, and it will be much more common to find rare variants that confer risk for bipolar illness or schizophrenia. It's also possible that there won't be any strong signals, that we're going to find maybe 20, 30, 40 variants, each of which has an effect size of 1.2, 1.3, and we'll need, then, to have 8,000 or 10,000 subjects to really see the risk architecture. In which case, you have to ask, well, is it because they're all acting in aggregate in some epistatic way to increase risk or is it just such a variable phenotype that it's hard to be able to relate any of these genes to a mixed bag of phenotypes?

This gets to your second question: would it be more likely to pick a highly restricted homogeneous sample and come up with a gene which gives you a greater effect and therefore would be easier to detect and might have more explanatory power? Maybe, but here's the problem: I don't think we know how to define such a sample. I think if you take any of these disorders, even autism, it's still not clear that the clinical phenotype as we see it will help us to define a homogeneous population at the genetic, the cellular, or the systems level. It might, but it actually hasn't played out that way for macular degeneration, it hasn't played out that way for hypertension, it hasn't played out that way for many disorders. So one spin-off from this, for instance, is to ask, should we do what was done for some of the degenerative disorders? Should you take childhood-onset schizophrenia and assume that they will have the highest genetic load and study that group separately? Some people are doing that and it may turn out to be useful. However, we could also find more genomic instability, a wider range of genetic lesions, not one consistent source, at the genetic level, for risk.

The final thing I'd say about all of this is you have to keep this in perspective. What we're talking about here is genetic contribution to risk, and we know that in people with identical DNA, that is, monozygotic twins, there's about a 50 percent recurrence rate, so the maximum you're going to find, no matter whether it's a genomic variant of huge effect size or many, many genes, all of which are studied in an epistatic fashion, is 50 percent of the risk. You may never get much beyond that at the population level. Genetics isn't going to tell us everything here; it should tell us some of the cellular pathways, it may identify some sub-syndromes, it could give us some new therapeutic candidates, but you've got to keep in perspective that, at the end of the day, it explains half the variance.

SRF: In this same vein, but at a higher level: there's a stereotype that U.S. research has the most money and funds the genetics and the biology. Other countries, for example, in the European Union, fund more of the research that might be relevant to experience effects or epigenetics. Could you talk a little bit about the differences between U.S. research and research in other countries and the things that we can learn from them that would bring the effects of the environment, writ large, to bear on the question of schizophrenia and psychiatric diseases?

TI: Well, I wouldn't accept the stereotype. At the level of environmental issues, a major set of challenges revolves around the lack of precision in the tools we have. We're hoping to be able to develop tools for epigenetic studies that will give us a lot more precision and essentially help us to say, that's where a footprint was for some environmental exposure prenatally, or at 6 months, or in the first year of life. We haven't had that. The European Union countries have put together this very interesting epigenetic consortium that's mostly focused on the mouse genome. They're already working hard to develop some of these tools, and we're playing catch-up a little bit. We are putting together an epigenetics program as part of the NIH Roadmap in 2008. Epigenetics is an area of future development, no question. We'll be doing a lot more in understanding not only what the environmental factors are but how they work. How do we find their effects after the fact, because that's often what we're left with, and it's one of the reasons why it's been so difficult to do this kind of work previously.

SRF: Could we partner with the EU or with the World Health Organization, either for that work or even for other "big" science such as WGA or other sorts of genomic research?

TI: For some of the genetics we certainly do that already. There's a great deal of opportunity here for collaboration, and some of that's already happening with Wellcome Trust and Canadian Institutes of Health Research.

I think for the World Health Organization you're talking about a very different organization. The WHO does not have large funds for research and they've supported very little research at the basic level. Much of what they are supporting is essentially demonstration projects to look at service delivery in resource-poor countries and to gather the information about services available in those countries. We currently collaborate with them in that effort. While we are working very closely with WHO on that end of the spectrum, that's pretty far away from the questions about mechanisms of schizophrenia and what we need to know to develop the next generation of treatments.

SRF: Let's shift gears and talk about the intramural program at NIMH. Can you describe the research on psychosis and related science being done at the institute, and some of the major findings that have come out recently?

TI: The intramural program exists largely for three reasons: to have a rapid response to an emerging public health issue; to develop resources for the extramural community to do innovative, high-risk translational studies; and for training the next generation of research leaders.

SRF: What's the scope of the intramural program? How many people, how big a budget?

TI: It's approximately 11 percent of our budget; we're spending roughly $160 million in the intramural program every year. We have over 1,000 people working in the intramural program, with 55 independent investigators who are tenured or on a tenure track. It's a relatively small group of investigators with a fairly large budget. The hope is that they do work that really can't be done anywhere else because it is high risk and would be difficult to make it through peer review; it's developing a resource, or it's a long-term project that just isn't likely to be supported easily extramurally.

Schizophrenia is a research focus of the intramural program as part of our Genetics of Cognition and Psychosis (GCAP) program under the leadership of Dan Weinberger. We have an in-patient service for long-term studies of medication-free patients. We have an active imaging effort in schizophrenia. And GCAP has a broad program to move from genetic findings into the realm of developmental neurobiology. The recent neuregulin paper, identifying a novel brain-specific isoform and localizing this protein in human fetal brain is a great example (Tan et al., 2007).

They've had four or five large projects around specific genes. The concept here is to move beyond statistical associations to try to understand what alleles in COMT, neuregulin, DISC1, or BDNF actually mean in terms of brain development and intermediate phenotypes. This requires a multidisciplinary approach, involving collaborators in several other NIH institutes, and incorporating a range of techniques from cell biology to neuroimaging to postmortem brain mapping. I don't know of a comparable effort in schizophrenia going on anywhere else.

SRF: In terms of clinical research, do you see any gains in research, intramural or extramural?

TI: The easiest way to define our goals in clinical research is in terms of our four "P's": personalized, predictive, preemptive, and pathophysiology. We've been talking about pathophysiology already, using genetics as a kind of gateway to understand what goes awry in brain development in schizophrenia, and imaging would be helpful on that as well. Bringing those two together with cell biology is going to be a brave new frontier for the field which I think will transform how we think about this illness. But the other three "P's" are where the more directed clinical research is going already. On Personalized, we are hoping our large-scale clinical efforts, like CATIE, will inform who does best on the available drugs. Obviously, current drugs are not sufficient. We have research looking at the next generation of drugs, targeting the greatest source of disability in schizophrenia: the cognitive deficits. You've already covered this in the Forum in various ways, and so you're pretty familiar with the kinds of projects that are underway—such as the MATRICS and TURNS efforts that we hope will give us the measurement tools and compounds that could be focused on cognitive deficits.

The Predictive/Preemptive part is in some ways even more transformative. We must identify the prodrome in its earliest stages to develop interventions that preempt the psychotic episode. I could imagine this having a real impact on the public health burden of schizophrenia and on the ultimate disability of the disorder. It's analogous to what we do now for coronary artery disease, where we predict risk at a very early stage, because of high cholesterol, family history, and sometimes changes on a thallium scan, and we preempt the worst outcomes by treating with statins, changing diets, and implementing exercise regimens. If someone has the first stages of angina, then we become even more aggressive. The result is that over a million deaths were averted last year for people who have coronary artery disease.

That's very much the model we ought to be thinking about for schizophrenia: predicting risk with genetics, detecting the prodrome with biomarkers, and intervening early, remembering that the psychotic phase of this illness may already be too late to have the best outcomes.

SRF: But the drugs that are available for schizophrenia are not as benign as the drugs that we have for cardiovascular disease, and there will be people with psychoses that are not schizophrenia, or young people who are not destined to develop schizophrenia who would be given our current drugs. Do you have any thoughts on that really tricky question?

TI: We're not giving the atypical antipsychotics or conventional antipsychotics at antipsychotic doses to people with the prodrome of this illness. Particularly because we're looking at a statistical risk and one can be certain that there'll be false positives. For preemptive interventions, you will need a high threshold for drug safety. We will need compounds that are much like the statins. And something perhaps even more important would be thinking about non-pharmacological interventions that could preempt the disorder. I don't think we've really thought through what those might look like. Some of them could be forms of social support, some of them could be cognitive tools that would be available on the Internet. There's a range of opportunities here that needs to be thought about when you're talking preemption as opposed to reversing a psychotic state.

SRF: My concern is that the cart would get in front of the horse, so that before those more benign drugs were available, the pressure of the profit impulse for pharmaceutical companies and the relative freedom that physicians have to control their own cases might lead to a dire situation for some people.

TI: I think everyone would agree that the current medications are not what we need. We need another generation of antipsychotics, for treating people with a psychotic episode, but even more so when it comes to thinking about the preemptive treatments. What's important to recognize is that in other areas of medicine, the preemptive interventions often look very different than those that you use after the disease progresses. So it's going to require stepping back and thinking about this more carefully. There has been some work already on cognitive behavioral therapy (CBT) as a preemptive intervention for the prodrome, published by Morrison and colleagues in the British Journal of Psychiatry a couple of years ago (Morrison et al., 2004).

SRF: I just noticed a new paper from the CATIE study showing that violence in schizophrenia has roots in pre-disease conduct problems (Swanson et al., 2007). It seemed like a fitting bookend, if you will, for a year that saw the death of Wayne Fenton and the killings at Virginia Tech, and so I wonder if we should be cautiously optimistic that a study like this could have some positive impact on efforts to combat stigma and perhaps focus research on root causes of violence and things like that. Any thoughts about that?

TI: Yes, it's a really interesting question; I like the way you posed it. In the past, my own sense of the community is that there's been an avoidance of talking about violence in schizophrenia because of a concern that it would increase the stigma. In fact, the data support the proposition that people with schizophrenia are more likely to be involved in violence either toward others or toward themselves unless they're treated. So there's every reason to treat people with this illness to reduce violence. But we also shouldn't avoid discussing the risk of violence in those who are not treated. Our failure to talk about the risk has led to loss of credibility in some of our attempts to reduce stigma. People with this illness, or with bipolar illness during the manic phase, are more likely to be violent than the general population by several-fold. Some people estimate that 50 percent of manic episodes involve violence, sometimes self-directed but other times not. What's interesting about this new study is that it breaks down the problem here; at least it clarifies that we may be talking about two quite different things. On the one hand, people who have a history of conduct disorder and who are at a greater risk for violence towards others, and for whom schizophrenia in some ways disinhibits them, exacerbates violent tendencies that they've had in the past. And that appears very different from people who are schizophrenic and are acting out a delusional fantasy.

As to the Wayne Fenton murder, in recent newspaper coverage of that, this young man said that he had a particular delusion about Dr. Fenton and he was trying to liberate Dr. Fenton's soul as a way of cleansing it. That's not sociopathy.

SRF: Can NIMH lead or at least encourage foundations or other groups in fighting stigma or increasing research for psychiatric funding?

TI: In terms of the need for funding for research, we all need to be very clear about NIH and NIMH as an investment rather than a cost. I think one of the things NIMH needs to do somewhat better is to be able to explain to the public what the costs of these illnesses are, and by that I mean more than the costs of treatment. We need to clarify the costs outside of the health care system: what we're spending in the criminal justice system, in homelessness, in social support such as SSI and SSDI. These indirect costs are enormous—they dwarf the costs of treatment. They're also largely in the public sector, so we're paying for this already. One of the things that I think will be very helpful is to get the public to understand that reducing this huge outlay of costs by coming up with new treatments, by coming up with biomarkers, coming up with better systems of care, is an investment against these very large outlays of expenses.

SRF: Any ideas about how to do that? Because it seems to me that the public, and not just the American public, always has a hard time appreciating those long-term benefits. They'll take the advantages of the SUV, for example, over worrying about global warming.

TI: Yes, I know there's some of that, but I think here we have an opportunity. In some ways the Virginia Tech massacre brings a lot of this to the fore: you can help people to understand what the costs are in the way that we're doing it now, and the costs in terms of lives lost, in terms of huge taxpayer burden. Much of that may not be sustainable. At NIMH we have started to recruit economists to help us understand these kinds of issues and help us frame them in such a way that the public can better understand them. It's surprising we haven't done this already, but in the past I think we were following a model which worked somewhat better in cancer research or in heart disease, where you can lay out what the costs are in terms of morbidity, mortality, and the financial expenditures for treatment.

SRF: Would your economists suggest something I've always wondered: joining NINDS and NIMH and NIAAA and NICHD and NIDA? These are all people studying the brain. Why not have some economy of scale where you save money? I'm reluctant to suggest putting a lot of hard-working public servants out on the street, but it seems like there's an argument to be made for having a "Brain Research Institute."

TI: We've been working collaboratively for a number of years on common issues, yet importantly keeping our distinct research missions. Most recently, we formed the Neuroscience Blueprint, which is a common pool for all those institutes and even a few others like the aging institute to come together to support such things as shared tool development, shared resource development, some shared training programs. We will spend $100 million in the first 5 years for this common effort. All the institute directors from these various institutes actually spent the day together at a retreat recently just thinking about what's the next best opportunity, what are the things we want to make sure we are supporting jointly. Areas like diversity training are so important and if we do this collectively, we will have much greater impact. And certainly as you just said, neuroscience, that is, the study of the brain, is relevant to all of these institutes, and it really doesn't make a huge difference whether you're studying synaptic plasticity in the cortex, the motor system, or the basal ganglia; you still have some of the same principles and they can be shared across institutes. We often use the example of looking at neurodegeneration in the retina as a model for neurodegeneration in the cortex, because we have a lot more access to the retina. So bringing these studies in NEI together with the relevant projects in NINDS has been a really helpful endeavor. Thus, the blueprint is actually an effort to do just what you're describing. Beyond the dollars, there is a shared intellectual vision and a collective effort to identify opportunities where we can raise all boats by working together.

SRF: Several years ago, NIH was pushing for open publishing or open access to journals, but I haven't seen a lot about that recently. I saw a statement from a group of academic publishers maybe a year and a half ago outlining their position about why this is not a good thing. Is there some movement on that front and what are the prospects?

TI: There's a great deal of movement on data sharing, which is different from sharing published articles, and that's happening for data from genomics, imaging, and epidemiology studies, as well as some clinical trials. The goal is to get all of the non-private data onto an accessible website so that people can do secondary analyses. As for published papers, PubMed Central was set up as a repository. About a year ago, when I last saw the outcome of that experiment, only about 4 percent of published articles were ending up in PubMed Central. In spite of that low figure, most scientists have access to papers through journal sites on the Web.

The bigger question is for the non-scientific community: how do they get access to publications? We're hearing two things. One, we're hearing, yes, we'd like to be able to get papers from the scientific journals more quickly than having to wait or having to pay for them, especially since we've already paid for the research that supported this work. The second thing that many people feel is that, well, it doesn't really help us to get some of these scientific articles because we can't understand them anyway. So there's a greater interest in reading some of the things that are publicly accessible from journals like the Annals of Internal Medicine and a few others that have begun publishing reader-friendly summaries of papers in the journal. That, I think, has had a little more pick-up from the community.

SRF: I think that was part of the success of the Alzheimer Research Forum: it was the first model where there was Web news coverage that was understandable to educated laypeople and people who are very interested in a particular disease area. Let me move to my "wrap-up" questions. Although you were entering a vast bureaucracy with many interest groups to contend with when you started as NIMH director, you probably nonetheless entered with a set of goals that you thought you could effect in terms of enhancing research, either intramural or extramural. What were those goals and how would you rate your success or achievements in the different areas?

TI: There were at least three areas that I came here to address, and I think we have made initial success on each of them. The first was to bring neuroscience to the study of mental illness. Clearly, that was overdue. And I guess the best way to say this is that if I looked back on the 20 years in which I was outside of the field of psychiatry � I left in about 1983, came back in 2002, 2003 � it was a little bit like being Rip van Winkle. I had not actually read the journals, I hadn't kept up in any way with research on mental illnesses, but I had been deeply involved in neuroscience, and I had been part of revolutions in molecular, cellular, and systems neuroscience during those 20 years. In neuroscience, it was a time of fantastic breakthroughs, with transformative discoveries at every level. Sometimes it seemed that there was almost nothing that we believed in 1983 that we still believed in neuroscience in 2003. At the same time, my experience was that in psychiatry in 2003, the field had not progressed; we were pretty much asking the same kinds of questions and using the same models that I had left in 1983. We had a few new technologies but no really transformative discoveries or insights, and if anything, the main difference was the degeneration of services for those with serious mental illness. It's not surprising that my first thought on arriving here was to ask, how do we bring the excitement and the revolutionary changes that have taken place in neuroscience and apply them to these brain disorders? So it was all about pathophysiology; bringing modern neurobiology to answer questions about schizophrenia, mood, and anxiety disorders. That required bringing in new investigators, it required making a very clear statement about priorities, and it required really shifting the emphasis in the Institute to mental illness. It was time to go where we could make the most progress. So that's number one.

Number two was to focus on the public health burden. It's great to bring neurobiology to these illnesses and to plan for biomarkers and new treatments, but we also had to find ways to optimize the treatments and services we have now and make a difference for the 20 million Americans who have these serious disorders today. I felt a real need to figure out a way to get beyond same old science: what would be the projects that could help us to do better with current treatments? How do we expedite what we do? Can we figure out who's going to respond to which drug? So that was part of what was the passion behind these practical trials like CATIE (see SRF related news story) and STAR*D, part of the reason for getting into pharmacogenomics, part of the reason for looking at biomarkers that would be predictive of treatment response, and recognizing that the treatments we have now really aren't the whole story, they're not where we want to be, but we've got to find a way to use what we've got better. So that was the second big area.

And then a third area, which is a little more process-oriented, was figuring out ways to integrate our intramural and extramural efforts. Intramural is where I grew up and it continues to make significant contributions to the field, but it's not well integrated with what we do in the other 89 percent of our budget. We have to think about ways to be able to optimize these two programs so that they work together, that they synergize in ways that we get the most out of both of them, and I'm still looking at that and trying to find the best way to do it. We need a new scientific director for our intramural program. We've been running with acting scientific directors for most of the last 3 years. Ironically, the science in that program has never been better, but we need some more permanent leadership and we need to bridge that program to the extramural community.

SRF: Why do they have to be integrated? If a solution to integrating intramural and extramural research hasn't become apparent, perhaps that's because there isn't one. If part of the purpose of the intramural program is to hire some smart, accomplished people to do research that's a little more risky, but with a potentially bigger pay-off, and the extramural program is for supporting the less risky research that funding panels can agree on, why shouldn't we as taxpayers support that division of efforts?

TI: Yes, integration may be difficult, but we need to make sure that we are getting the most from both efforts: synergy may be more efficient than separation.

SRF: So that last goal would be a work in progress, and the other two goals, would you say you had achieved them at some level?

TI: We're achieving all of them. Again, we're in process, but I think we're making real progress on all fronts. I'm tremendously proud of the intramural program and where the work is going there. I have to say if you just look at the recent studies on depression, bipolar illness, and schizophrenia and this whole new area of imaging genomics, that program has really been important for our discovery base. On the other hand, it still isn't well integrated, so I think we've got a lot of good stories to tell, but are still not where we need to be. I'm certainly not finished with what I've set out to do here; there are another few years of hard work to get us further down the path.

SRF: Well that was going to be my last question. Now that you've been in public health for a few years, where do you see yourself in 5 years? Still at NIMH or back in a lab studying the basis of social behavior in a rodent model?

TI: I think once you've been away from the lab for 5 years, it's harder and harder to go back. I'm a science junkie at heart, so I read everything I can get my hands on that has to do with neuroscience and mental illness. I keep up with as much science as I possibly can. But I think I'm not going to be a bench scientist in the future. I've become much more interested in trying to make a difference at the level of public policy and administering research efforts. That's probably in everybody's best interest. I feel like there are many talented people who can do the lab work so much better than I can at this point. Though I've got to confess, there are days when I really miss it. It's partly that sense of discovery, but most of all it's just the culture of being in a scientific group where you have students and a sense of family that you miss when you're in a public policy arena.

SRF: Is there anything else that you thought should have been brought up?

TI: Just one thing. Frequently people ask me, "Gosh, why would you want to do this job? You picked the right job but the wrong time." That's often the way the question comes up because of the budget. In fact, I feel like there's never been a better time to be here because at the end of the day, this is a scientific job. It's really about finding the very best science that's out there and applying it to those illnesses that cause the most disability. We've never had so much great science to pull from. To me, it's pretty much unprecedented to realize that every issue of the major science journals in neuroscience is taking us further down the path where we want to go. It's moving very quickly right now, so, from where I sit, there's never been a better time to be here.

SRF: Great. Thanks so much.

TI: Take care.