| Preskorn.com | Printed from: http://www.preskorn.com/columns/0111.html |
| Drug Development in Psychiatry and Genomics: From E. Coli to Man |
|
|
|
SHELDON H. PRESKORN, MD |
|
Journal of Psychiatric Practice, November 2001, 415-419 |
In the last column, the paradigm of anti-infective drug development and its application to psychiatric drug development was discussed. Protease inhibitors were used to illustrate how drug development can be targeted to specific sites of action. Azole antifungals were used to illustrate how unintended affects can be produced due to the fact that even yeast and man share some genetic similarities.
This column will continue this theme by discussing how the knowledge gained from the sequencing of the Escherichia coli (E. coli) genome is likely to affect psychiatric drug development. The column will also expand the discussion begun in the last column concerning the greater challenges encountered in developing drugs to treat psychiatric illnesses versus infectious diseases (Table 1), even while acknowledging the commonality between these two development processes. Understanding these challenges is conceptually important for both the prescribes and the researcher. This knowledge provides a context for understanding the limitations of current psychiatric drug development as well as how the human genome project will likely change this process in the future. To paraphrase Steinbeck, this column could be entitled "Of E. coli and men."
Potential Targets
The number of genes in the human genome has still not been determined with certainty; however, the latest best estimates are 40,000-50,000 genes. In contrast, the E. coli genome contains approximately ten times fewer genes: 4,401.1 Yet, the sequencing of the complete genome for this organism, which was first accomplished in 1997 by two different groups of scientists,2,3 has considerable import for our understanding of the cellular paradigm (i.e., the mechanisms of life at a cellular level) and, hence, for drug development in psychiatry.
The importance of E. coli was underscored by Schaechter and Neidhart, who stated: "All cell biologists have two cells of interest: the one they are studying and Escherichia coli."4 This is true even for the cell biologist interested in brain function, even though all the cells in even the most primitive brains are eukarocytes (i.e., have a true nucleus bounded by a membrane), whereas E.coli is a prokarocyte (i.e., an organism with its DNA or nuclear material scattered in the cytoplasm of the cell but that nevertheless reproduces by cell division).
The E. coli cell achieved its vaunted position in the late 1930s when the forefathers of modern cell biology concluded that it was only by elucidating the simplest biological systems that the relationship between genes and cell function could be understood. The goal was to understand the logic of life based on the concept that the mechanisms of life are finite, tractable, and subject to full exploration and understanding.
The E. coli cell became a cornerstone in this undertaking for a number of reasons4,5 Its simple nutritional requirements meant that it could be isolated and grown relatively easily in the laboratory. A new generation was produced every 20 minutes, a feature that allowed study of the effects of mutations on the synthesis or use of essential metabolites. It was generally well behaved, with a low risk of pathogenicity, making it relatively safe for laboratory study. It was a sexual organism in that it demonstrated the ability for genetic recombination between cells. All of these features facilitated study of the linkage between biochemistry and genes.
| Table 1. The differences between developing antibiotics and psychiatric medications | ||
| Item | Antibiotics | Psychiatric medications |
| Number of genes coding for targets of potential relevance to drug developers | 500-4,000 | 40-50,000 |
| Treatment goal | Interrupt normal cell biology | Restore normal cell biology |
| Knowledge of clinically relevant mechanism of action | Based on disease process | Based on action of earlier drugs |
| Accuracy of preclinical predictions | High | Low, except for "me too" drugs |
| Inclusion criteria for clinical trial | Specific to the disease process | Enrollment based on having a clinical syndrome |
| Study endpoints, both preclinically and clinically | Simple to measure, dichotomoous, parametric, unambiguous clinical relevance | Difficult to measure, dimensional, nonparametric, debatable clinical relevance |
As a result of the extensive study of E. coli during the last 60-70 years of the 20th century, more was (and probably still is) known about the biochemistry of E. coli than of any other living organism by the time its genome was sequenced. E. coli was found to have a relatively complete repertoire of the biochemical processes associated with more advanced forms of life, including the ability to use glucose and inorganic salts to synthesize virtually everything needed for life. E. coli was also found to be a highly adaptive and versatile organism capable of using an impressive array of organic compounds as a source of nitrogen and carbon. This latter feature further enhanced its stature in cell biology by making it an ideal model organism in which to study gene regulation and adaptive evolution.
Nevertheless, E. coli was not the first, but rather the seventh, bacterial genome to be fully sequenced. The others included Haemophilus influenzae with 1,727 genes and mycoplasma genitalium with 470 genes.6 In fact, the latter is the smallest and simplest self-replicating organism capable of independent life in existence and has a circular double-stranded DNA typical of prokarocytes. For these reasons, Morowitz, in 1984, proposed using mycoplasmas as models for defining the entire machinery of the basic living cell in molecular terms.7 Thus, the study of mycoplasmas and E. coli has formed the foundation of modern cellular biology.
The work on the biochemistry and genetics of E. coli over the last 70 years has been highly fruitful and included the elucidation of major biosynthetic pathways, the mechanisms underlying gene regulation, transcription, and translation, and the fundamentals of DNA replication, mutation, and repair. These findings are relevant to the understanding of life at a cellular level. Despite the importance of this work, the sequencing of the entire E. coli genome revealed that 40% of its genes had no known function.2 The decades of conventional biochemical and genetic study had identified perhaps 80% of the genes required for the biochemical and regulatory pathways necessary for the normal life of E. coli. Yet there were many more genes whose functions had escaped discovery. The sequencing of the E. coli genome both made their existence known and provided ways to elucidate their function. This fact underscores the importance and the implications of the human genome project, because the number of human genes whose function is still unknown far outstrips that of the E. coli.
The E. coli genes with no known function or homologies were labeled "function unknown" (FUN) genes in the sequence databases. They provided cell biologists with entirely new targets to investigate and understand as well as the means to accomplish this goal.8 These genes were likely to subserve more complex function such as
In the 4 years since the sequencing of the E. coli genome was completed, there has been an explosion in knowledge about the role of these FUN genes in the life of the E. coli and thus a more complete understanding has been gained of gene regulation at the level of an intact, living, and reproducing cell with all of the basic biochemical and genetic repertoire for independent life. These advances in knowledge of cell biology coupled with advances in information technology have permitted the development of publicly accessible databases that allow scientists around the world to both use and expand on this information.1,9 For the interested reader, such databases include:
Underscoring the importance of the study of E. coli and its relevance to psychiatric drug development, only 7% of the protein-coding sequences for E. coli are unique to that organism.2 At the time of its sequencing, the E. coli genome was found to contain 158 genes that were homologous to other known genes and an additional 232 that were identical or similar to hypothetical genes recorded in other genomic databases.3 In fact, the E. coli genome contains three genes homologous to human genes that code the long-chain fatty acid degradation enzyme complex in the mitochondrial membrane. The E. coli genome codes for over 695 enzymes that mediate over 595 metabolic reactions organized into 123 metabolic pathways.10
A similar but even more extensive and protracted explosion in knowledge can be expected to result from the sequencing of the human genome. This will lead to an increased knowledge of normal human cell biology and of abnormalities in that biology that result in disease, including psychiatric illnesses. This will, in turn, lead to the discovery of more "high yield" targets for drug discovery relevant to psychiatric medications. These discoveries will also help researchers address the other challenges facing psychiatric drug discovery (Table 1).
Treatment Goal
Drug discovery for anti-infective agents has a much simpler goal than do most other areas of drug discovery, including psychiatric drug discovery. The goal in developing anti-infectives is to interrupt normal cell biology so that the target organism is not able to either sustain or reproduce itself. The common mechanism is to produce a drug capable of inhibiting a mechanism (i.e., an enzyme or ribosome) vital to a biochemical pathway essential to either maintenance of cellular integrity ("cidal" drugs) or reproductive function ("static" drugs). Once a critical mechanism in an essential pathway has been identified, then the task is to find a molecule that can block that mechanism and yet be given safely as a drug to a human. As mentioned in the preceding column, that is made simpler by virtue of the fact that the target mechanism is alien and thus may differ substantially from vital human mechanisms.11 As discussed above, there is nevertheless the problem of conservation of genes or at least homologues of genes across species even as disparate as E. coli versus human.
In contrast to antibiotics, the goal of most other types of medications is to restore normal cell biology. That goal requires two kinds of knowledge:
Finding answers to these questions makes the task difficult enough, but there is also the problem of how to correct the problem once it is identified. The problem with restoring normal cell biology is that one must replace or compensate for something that is not working correctly. That is considerably more difficult than blocking something that is working correctly. If the problem is a deficiency in a normal product (i.e., insulin), then the treatment can be relatively simple (i.e., giving the deficient substance). If the problem is a defect in translation (e.g., the genetic defects suspected of underlying some forms of Alzheimer's disease), then a simple molecule such as a drug may not be capable of correcting the problem--hence, the interest in gene therapy.
Clinically Relevant Mechanism of Action
The preceding discussion is directly relevant to differences in our knowledge of what mechanisms are likely to produce a desired clinical response. Recall the equation central to these columns:
Effect = affinity for x drug x biological
site of action concentration variance
(Equation 1)
The mechanisms essential to the normal cell biology of the infectious organism are the potential sites of action for the development of anti-infectives. The degree to which those sites of action differ in their structure from sites of action in humans defines the therapeutic index of the drugs (i.e., the dose that will be effective versus the dose that will be toxic [variable 2 in the equation]). Certainty about the desired target is a function of knowledge of either the disease process or of other drugs that have proven to be clinically useful in the same condition. In the case of anti-infectives, the goal is to eliminate the infectious agent causing the disease. The recognition that infectious agents could cause disease and the need to determine the identity of those agents was the logic underlying Koch's postulate established over a century ago.
The development of psychiatric medications has until now followed a pattern involving, first, the chance discovery of clinically useful agents, followed by the development of newer refined drugs based on a knowledge of the pharmacology of the earlier chance discovery drugs. This topic was the focus of an earlier series of columns on the development of antidepressants12 and of the first column on understanding the pharmacology of antipsychotics.13 While this approach has been quite productive in terms of producing more "user-friendly" and "patient-friendly" medications, such as the selective serotonin reuptake inhibitors, it leads to the development of drugs with the same mechanisms of action and hence the same spectrum of activity rather than to the development of novel compounds that have the potential to work in patients who do not respond to existing medications.
The unraveling of the cell biology relevant to psychiatric illnesses is the great hope that the human genome project holds for psychiatry. This knowledge will lead to the identification of novel targets (the first variable in equation 1). That process in turn will be advanced by the understanding of simple life forms such as E. coli.
Accuracy of Preclinical Predictions
Recall that drug development is a gamble.14 However, the more one knows about the basic biology of a disease, the more one reduces the gamble, because it is easier to identify high probability targets. Similarly, the more one can model the disease process in preclinical systems, including single cell organisms that may even have been "humanized" to express the target of interest, the more one reduces the gamble. This is important because preclinical testing is much faster, cheaper, and less risky than clinical testing. In fact, the more one knows about the relevant biology, the more clinical testing becomes simply a "proof of concept." Recall that perhaps the most cogent goal of drug development is to reduce uncertainty about the effects of a drug. The more one knows about the relevant mechanisms to target and to avoid, the more one has already reduced the uncertainty about the drug's potential effect.
Identifying targets to hit and to avoid is the essence and promise of the human genome project. Findings from that project, coupled with the technique of high throughput screening, permit rapid identification of the critical structure needed for the molecule to affect a specific target while avoiding other targets. This process will advance new drug discovery and reduce the inherent risk for drug developers by enabling them to avoid problems such as were encountered with drugs like fluoxetine and paroxetine, which inadvertently inhibited enzymes responsible for oxidative drug metabolism in humans while also blocking serotonin uptake, their desired mechanism of action.14
Inclusion Criteria for Clinical Trials
If the biology underlying the disease process is known, as is the case with infectious diseases, then the inclusion criteria can be quite specific (i.e., the isolation of the offending organism-one of Koch's postulates). This results in fewer false positives and hence enhances the signal-to-noise ratio.
However, imagine trying to determine whether a highly specific antibiotic is clinically effective in a patient with pneumonia without being able to determine whether the patient has a viral or bacterial infection, much less what specific type of virus or bacteria is involved. Yet this is close to the situation facing the developers of psychiatric medications. Inclusion criteria for psychiatric clinical trials are based on having a psychiatric syndrome without knowing whether all individuals who meet that syndromic diagnosis have the same underlying biology-and the prevailing evidence from many different perspectives suggests that they do not.
This problem likely contributes substantially to the small signal-to noise ratio plaguing psychiatric drug development.15,16 Specifically, it requires the enrollment of more subjects to test whether the compound is effective, which in turn costs more money and takes more time. This longer development time cuts into the patent life of the drug, which is how the -sponsor achieves a return on investment. A small signal-to-noise ratio also increases the likelihood of false negative studies. For example, even the most widely used antidepressants that form the cornerstone of treatment were only found superior to placebo in approximately half of the studies in which they were tested. Thus, in antidepressant drug development, one needs to do two studies to get one positive result, making psychiatric drug development both more expensive and more risky.
Study Endpoints
Another problem that has resulted from our lack of knowledge about the relevant biology underlying psychiatric illness is determining what to measure in clinical trials of psychiatric medications. The endpoint measures commonly used in psychiatric clinical trials are rating scales based on symptoms of the syndrome, such as the Hamilton Depression Rating Scale (HDRS)17 and the Positive and Negative Syndrome Scale (PANSS) for schizophrenia.18
There are many problems with such scales from both a scientific and a statistical perspective. From a scientific perspective, they are dependent primarily on face validity, and there is no certainty that they are measuring fundamental aspects of the biology of the illness. For example, the development of the HDRS was based as much on the pharmacology of tricyclic antidepressants as on the symptoms of major depression. As such, this scale favors drugs that are sedating and cause weight gain. This is because 3 of the items on the most widely used 17-item version of the HDRS ask about initial, middle, and terminal insomnia, while 2 other items rate appetite and weight gain. Moreover, these items are weighted more heavily than items such as depressed mood, suicidal ideation, or psychomotor retardation. For example, the three items for sleep total 6 points compared to 4 points for depressed mood and suicide ideation.
This fact handicaps drugs such as the SSRIs that are not sedating and do not cause weight gain, even though the SSRIs are now the most widely used antidepressants. In fact, this argument was unsuccessfully advanced to explain the inability to demonstrate antidepressant activity for fluvoxamine.19
In terms of face validity as well as the appropriateness of inclusion criteria, there is the question of what to do with so-called atypical depression, in which the patient has hypersomnia rather than insomnia and hyperphagia rather than anorexia. The usual inclusion criteria for antidepressant clinical trials require a minimum score of 20 on a 17-item HDRS. Although such criteria do not exclude patients with atypical depression, they make it more difficult for such patients to qualify. The absence of items that reflect a drug's effect on these symptoms on the 17-item HDRS also makes it more difficult for drugs that improve these symptoms to be found effective in trials that utilize the HDRS as the primary efficacy measure, and thus biases against finding drugs that may be particularly effective for patients with atypical depression. Contrast this situation with trials investigating treatments for infectious diseases, in which the outcome endpoints can be tied to the disease process (i.e., the elimination of the causative agent).
Obviously, many diseases are not as straightforward as infectious diseases but are nevertheless better understood in terms of their underlying biology than is currently the case for psychiatric illnesses. Examples include atherosclerosis and inflammatory processes such as osteoarthritis, in which cyclooxygensase II is an important mechanism. In such diseases, surrogate markers of the imputed disease process (e.g., reductions in the levels of specific types of lipids in the blood or in the titers of specific immunoreactive substances) can be used to test the effect of the drug. Such surrogate markers may become available for specific neuropsychiatric disorders in the near future as advances are made in our understanding of the biology underlying certain illnesses (e.g., amyloid production and deposition in specific types of Alzheimer's disease). Such advances are important, first, because such outcome measures are more quantitative and less open to varying interpretation, and also because such markers can be identified earlier in the disease process, which may take years to become clinically apparent. Such a long time span makes clinical trials for drug development essentially impossible from a financial perspective, because the patent could lapse before sufficient evidence could be gathered to prove that the treatment is effective in preventing or delaying the onset of a disease (e.g., Alzheimer's disease or atherosclerosis).
This column thus provides a bridge between the problems inherent in drug discovery in psychiatry and the challenges inherent in testing such drugs. The problems inherent in human testing will be the subject of future columns in this series.