Editor’s Note: The evening session on Monday, 9 March, was the final hurrah of the Keystone meeting and was devoted to “Chemical Biology and Therapeutic Strategies” (a title that led session organizer Christopher Ross to quip that chemical biology is really just pharmacology). We are indebted to Bartlomiej Lukasz, of University College Dublin, Ireland, winner of one of the travel scholarships funded by NIMH, who filed this meeting report on the two summarizing talks of the symposium, as well as some notes on the general discussion that followed.
3 April 2009. In the first talk, John R. Kelsoe, University of California, San Diego, presented drug response as a promising path to better therapies, and also as an endophenotype that could lead to better understanding of underlying mood and cognitive disorders. In bipolar disorder (BP), there is a limited or incomplete response in 30-50 percent of patients (Kennedy and Lam, 2003), which is certainly due in part to a lack of knowledge of pathophysiology, said Kelsoe. But in addition, diagnosis and treatment are hampered by the limited predictive value of diagnoses, limited longitudinal stability, and limited predictability of treatment response. This leads to trial-and-error treatment. One study found a 69 percent misdiagnosis rate among 600 surveyed bipolar patients, with over one-third waiting 10 or more years and visiting several physicians before being diagnosed properly (Hirschfeld et al., 2003).
Kelsoe focused his talk on lithium, the oldest, but still the best BP drug, in his opinion. It reduces suicide risk and virtually “cures” some patients, but has largely been supplanted by valproic acid in the United States, something Kelsoe deplores. He described a series of interesting studies, revealing specific SNPs in the NTRK2 gene associated with a euphoric mania sub-population of lithium-responsive bipolar patients (Bremer et al., 2007). This gene codes for the BDNF receptor TrkB; expression of BDNF is necessary for lithium and other antidepressants to generate a response in animal models.
Based on these findings, he suggested that individual differences in drug responsiveness may mark biologically different entities currently perceived as one disease. Kelso suggested that further studies of genetic variants linked to specific drug response profiles in the context of disease susceptibility genes could give us more in-depth knowledge about disease pathology. Treating these drug response profiles as specific endophenotypes could also help in developing tailor-made treatments. Kelso concluded with the hope that the use of genetic screening in clinical practice will bring better therapies based on disease mechanisms, more accurate prognoses, and treatment outcome predictions.
In the second talk, Edward M. Scolnick of the Broad Institute of MIT and Harvard opened with the reminder that the last 50 years have not brought any mechanistically new drugs for either schizophrenia or BP. Current antipsychotics, despite being vastly improved over phenothiazines, still target multiple dopamine, serotonin, and acetylcholine receptors, resulting in complex responses and serious side effects (Scolnick, 2006). This lack of specificity for one molecular target, in contrast to most non-psychiatric drugs, seriously limits the possibilities of fine-tuning therapy. Concerning BP therapies, the problem is a lack of a consensus about the mechanisms and pathways beyond their therapeutic effects. Scolnick pointed out that, in comparison to therapeutic developments in other diseases, psychoactive drugs often skip some parts of a drug discovery pipeline. This does not make deciphering the underlying mechanism any easier, and we should all learn from the work leading to mGluR5 antagonist candidate drugs for fragile X syndrome development (Bear et al., 2004; see also SRF related Keystone story).
Scolnick suggested that autism is a potential model for schizophrenia and BP in terms of research (though he acknowledged that there was a bit of luck to have found Mendelian mutations in autism). These findings may allow researchers to employ mouse models and cell assays to identify molecules that could reverse the disease gene-related phenotypes in these model systems. This might, in turn, lead to human trials in the genetically defined variant, and even in the broader disease. Echoing many at the symposium, Scolnick was particularly excited about the potential of induced pluripotent stem cells, which can be used to create neuronal cultures derived from skin cells—and containing the genetic backgrounds—of individuals with psychiatric disorders. Even without Mendelian mutations, Scolnick suggested that GWAS of schizophrenia and BP, together with other studies such as the 1000 Genomes Project, should be able to identify disease-related SNPs of medium penetrance and medium commonness that could fuel similar research strategies.
What I really enjoyed in this talk was Scolnick’s reminder about environmental influences. Both schizophrenia and BP are complex diseases, where probably multiple genetic variants (SNPs, CNVs) and environmental factors interact, and despite immense progress in the field of whole genome analysis, we still have a lot to discover. He also made the point that work on Fragile X and other disorders considers the possibility of rescuing even neurodevelopmental lesions.
At the end of his talk, Scolnick stressed that further studies of the wnt signalling pathway (regulated by lithium and DISC1) may be our best chance for new therapeutic targets in schizophrenia and BD. Other promising compounds, according to Scolnick, include mGluR2 agonists, N-desmethylclozapine, and nicotinic and GABAergic drugs, as well as a novel regimen of D-cycloserine treatment combined with psychotherapy (Otto et al., 2007) and phosphodiesterase 10 inhibitors (Schmidt et al., 2008).
The symposium finished with general discussion opened by Judith Rapoport of NIMH, who asked how decisions should be made about when to proceed with regular screening/opening of the drug development pipeline for new, promising compounds under study. No clear and full answer was given to this difficult problem; however, it is worth mentioning Scolnick’s announcement during his lecture about financial support, know-how, and facilities available at the Broad Institute for developing screening strategies for novel neurotherapeutics. The next question from the audience raised the problem of phenotypes associated with, or that develop during, progression of mental disorders, e.g., epilepsy accompanying schizophrenia or migraine in case of BP. The organizers agreed that more attention should be given to systematic and quantitative analysis of comorbidities.
A comment, from an attendee representing a pharmaceutical company, concerning the limited usefulness of GWAS for translational approach and drug development kindled further discussion. Against his point about often observed discrepancies between phenotypic screens and genotype (the role of compensatory mechanisms was mentioned), the geneticists replied that GWAS should be treated more as a powerful ally in basic research. Pamela Sklar of the Broad Institute addressed the question of why such large samples are needed for GWAS: she said that “ironclad” evidence was needed given the complexity of the biology, and noted that this approach was likely to lead to “generalizable” genes, at work in more than just one population. These studies bear potential for new and “eye opening” findings; however, many more dots remain to be connected and sometimes less stringent “p-value control” considered. This side of the argument was significantly strengthened by Daniel Weinberger, NIMH, who stressed that high-density SNP analysis can reveal more than previously thought. He also pointed out that classic genetic tools may not be suitable for mental diseases, where complexity seems to be the key. In this stochastic mixture of the changes at genetic, pathway, and protein levels, we need solutions which will cover and connect fields of genetic analysis and diagnostics.—Bartlomiej Lukasz (with additional reporting by Hakon Heimer).