Antidepressant interactions overview
A Brief Review of drug-drug interactions involving antidepressants
Introduction and background
A large proportion of patients do not get an ideal result form the first antidepressant that they are given, and it is frequently necessary to consider alternatives which may involve combinations of drugs. Experience indicates that a majority of patients being treated for depression take more than one drug, for one indication or another. At present the fashion of using selective serotonin reuptake inhibitors (SSRIs) endures, and, because this family of drugs have a higher chance of interacting with other drugs, drug-drug interaction problems are common and sometimes serious. Although selective serotonin reuptake inhibitors (SSRIs) have been in use for two decades many doctors prescribing them are still at an early stage of developing knowledge and understanding of even the most serious and dangerous interactions. It is not at all uncommon for me to see patients who have been prescribed problematic combinations (by fellow specialists). I am unable to explain why this is so, or why the specialist colleges continue to award doctors the imprimatur of their membership when even recently qualified members prescribe dangerous combinations and where a significant proportion of all of them remain at such a low level of knowledge of pharmacology. In my opinion patients and anyone else, especially doctors, who wish to check they do not have combinations that are potentially problematic (or even dangerous) should make themselves familiar with this web site and flockharts (make it a Web browser favourite) and get into the habit of checking everything. http://medicine.iupui.edu/flockhart/table.htm. The information on this (my) website has more background and explanation and hopefully will be helpful for many people. If you cannot spot the correct response in the quiz questions in this section you will undoubtedly benefit from increasing your knowledge base. If it looks like there may be an interaction ask the difficult questions and if satisfactory answers are not forthcoming, consider your options carefully, and think about a second opinion.
In order to make a balanced decision concerning treatment benefit vs. interactions risks it is useful to recap on some relevant factors and perspectives. Accordingly, the brief comments that follow start with information intended to frame and enable that process.
Many meta-analyses have been performed comparing the various antidepressants, but no clear consensus concerning any superiority of one over another has emerged: that is because the underlying quality of the scientific evidence is insufficiently good. Selective serotonin reuptake inhibitors (SSRIs) and other non tricyclic antidepressant (TCA) drugs are very probably less effective than either tricyclic antidepressants (TCA) or monoamine oxidase inhibitors (MAOI) or ECT for severe depression. The data on newer drugs has not yet been independently replicated. Pharmaceutical company sponsorship creates a bias that inflates their apparent efficacy advantage, they also get disproportionately promoted to doctors. SSRIs may have marginally less side effects than tricyclic antidepressants (TCA), an NNT of 33* (1). A number needed to treat (NNT) means that you would need to treat 33 patients with selective serotonin reuptake inhibitors (SSRIs), instead of a tricyclic antidepressant (TCA), in order to finish up with one extra patient who is able to tolerate the treatment by virtue of experiencing less side effects. Even taken at face value that is hardly a quantum leap forwards. That statistic is made less valid when one appreciates that it is illogical to make comparisons between groups of drugs. It is meaningful to compare one drug directly with another, e.g. nortriptyline with fluvoxamine, but not groups of drugs. You cant take an average SSRI or TCA, you can only take one particular drug. What is also important, but seems not to be appreciated, is that the greater the need for effective treatment, the less the advantage of SSRIs is. This is because the data is derived from patient samples heavily weighed with subjects with mild illnesses, and / or those who do not really have depression at all, at least not drug responsive depression. As dear old Winston Churchill is reputed to have said, lies, damned lies, and statistics.
Gillmans maxim no. 12 The difference between advertising and statistics is that advertising is seen mostly on TV, and statistics mostly in scientific journals; the similarity is they both assist in the generation of profits.
Clinicians often use combinations of antidepressants, in Australia about 5% of SSRI scripts appear to be co-prescribing with TCAs (2). There has always been an official view from professional colleges, and most guidelines, that combinations are dangerous and should not be used; the recent Australian guidelines specifically condemn this practice (3). It is clear that a general understanding of drug-drug interaction problems among doctors is poor (4). It is difficult to decide what conclusion this should lead to in relation to medical education. Posterity suggests that a short account of a meeting I had with two eminent professors of Psychiatry (psychopharmacology) in the UK in the early 1990s would be in place here. I was eager to learn of their experience of dual-action strategies by mixing SSRIs with NRIs, to see if it was as positive as mine (a little reality check for me, especially in view of my isolation in Tropical North Queensland). The reaction was alarm, indicating that it was dangerous and that I should not do it. My heart missed a beat, and I immediately asked what terrible problem they had encountered that I did not know about. After all, I was in the middle of nowhere in North Queensland, a bit out of touch perhaps. The answer was, of course, that CYP450 interactions would cause elevated blood levels death mayhem etc. I smiled with relief. I had already published on this observation from one of my early cases involving that difficult drug fluoxetine, in 1990, one of the earliest such reports, and the first outside the USA (5).
I responded to these two eminent professors and suggested that you did not have to be a rocket scientist to understand and manage such problems. They both indicated that the average UK psychiatrist, in their judgement, was not capable of achieving that, and was not rocket scientist material. They indicated that their policy was a dogmatic blanket prohibition of any such combination. My response was, and still is, that this was an appalling condemnation of both academic and clinical psychiatry. That was a significant formative experience underlying my decision to call myself as a clinical neuropharmacologist, rather than a psychiatrist. It reminded me of how I nearly changed specialties early in my career. I was concerned and about poor science and hocus-pocus in psychiatry, as well as the second-rate calibre of entrants into that field. It was certainly a second choice, or worse, for many people in my era. It was clear that the field had more than its share of intellectual fraud. That could be a slip of the tongue, and mean also, too full of Freud. I subscribe to the view so splendidly expressed by the Nobel prize winner, and great communicator, Sir Peter Medawar (6) who said, Opinion is gaining ground that doctrinaire psychoanalytic theory is the most stupendous intellectual confidence trick of the twentieth century: and a terminal product as well – something akin to a dinosaur or a Zeppelin in the history of ideas, a vast structure of radically unsound design with no posterity. I do wish I had known this quote at the time I refused to participate in psychoanalysis tutorials. I felt quite proud when I read, later, what Sir Peter had written, especially because I used similar words the biggest intellectual confidence trick of the twentieth century when stating my refusal.
A significant percentage of hospital admissions throughout the Western world, approaching 10 per cent, are estimated to be primarily caused by drug interactions (7,8). This is ironic and related to the diversion above, because, throughout the previous four decades it was considered by most that there was little point in measuring blood levels of antidepressants. I am quite sure many of these drug related admissions were of poor metabolisers who were over-dosed by their doctors. I certainly saw a few myself, and indeed still do. Few people seemed to be concerned that it was a significant problem. Now that we are able to create poor metabolisers by blocking CYP450 with problematic antidepressants like fluvoxamine, fluoxetine and paroxetine etc. everyone seems to have gone to the other extreme of getting irrational and emotional about the dangers of combinations. But, for all those years, the natural poor metabolisers (genotypic PMs) were in just as much danger as the drug-drug interaction created poor metabolisers (phenotypic PMs). Such is the (enduring) legacy of ignorance. It underlines my points above, and I remain apologetic for the psychiatry. Key opinion leaders (KOLs) continue to jet around the world pontificating about the mostly imaginary, and always exaggerated, advantages of newer drugs, whilst basic post-graduate education remains a poor relative (see (9)).
Information concerning the interaction potential of TCAs is deficient because there has been no comprehensive attempt to analyse their capacity to inhibit CYP450 enzymes, or, for that matter, ion channels or receptors. The reason for this is presumably related to their insignificant commercial value, coupled with the fact that research money for non specific interesting things not related to commercial gain is thin on the ground. Although such properties of the SSRIs are a little better documented the information is not complete or well disseminated because there is a nebulous obligation on pharmaceutical companies to update their product information, at least not as assiduously as they detail doctors concerning the supposed advantages of their drugs (9). Not only is there no obligation for pharmaceutical companies to update information, but also there is a very specific and direct disincentive to open any potential cans of worms. If your drug is already on the market, and then some interesting new development, like an increased understanding of CYP450 comes along, you do not want to ask questions you dont already know the answer to: the answer might turn round and bite you. Unless a competitor claims your drug is worse, in some particular respect, than theirs, its wise to keep your head below the battlements.
Pharmacists have an important role to play in ensuring that adverse drug reactions are avoided, and appropriately reported if they occur. They also share a legal and professional responsibility towards the care of patients.
Pharmaco-dynamic interactions occur when the end effects of two drugs interact in a meaningful way to alter side effects (or beneficial effects). An example of this is the increase of brain serotonin by simultaneous blockade of monoamine-oxidase (MAO) and serotonin reuptake (SRI). This is the mechanism by which serotonin toxicity, serotonin syndrome, can cause fatalities. It may be noted in passing that pharmaco-dynamic interactions tend to be more severe than pharmaco-kinetic ones. There is a good reason for this: if you think you know what it is, you may email me, and I will post the answer and name of the admirable intellect who produces it. If this answer conflicts with what I think is the correct one, an interesting debate may ensue.
The other mechanism is referred to as pharmaco-kinetic, because one drug interferes with the metabolism of another in a way which alters its concentration at the relevant site, and therefore its magnitude of action. Such interactions effectively alter (often increase) the level of the other drug. That is what occurs when problematic SSRIs like fluoxetine, fluvoxamine and paroxetine interfere with the breakdown (inactivation) of many other common medications, as they so frequently do. The difficulties in this area are compounded by the fact that there is much obfuscation and mis-information, in pharmaceutical company product information texts, and even in supposedly authoritative reference texts such as the British National Formulary and the Australian medicines handbook, especially concerning interactions of TCAs with MAOIs (see Gillman 2006 (10)).
The problem with combining MAOIs with re-uptake inhibitors (i.e. either SSRIs or TCAs) is that of serotonin toxicity (serotonin syndrome). This reaction can only occur with antidepressants that posses potency as SRIs. Of the TCAs, this only includes imipramine and clomipramine. The other TCAs are safe to combine with the MAOIs because they do not posses significant SRI potency.
TCAs are less troublesome than the SSRIs in relation to CYP450 interactions. Indeed, two of the SSRIs, fluoxetine and fluvoxamine, would likely not be approved today because of CYP450 interactions. They should be used cautiously, if at all (9,11).
It is necessary to be informed about these phamacokinetic interactions in order to avoid problematic combinations, and to make compensatory dosage adjustments. Such dosage adjustments are also required in patients not in danger of drug-drug interactions, because of the significant proportion of genotypic slow (poor), and rapid, metabolisers in all populations For instance, poor (slow) metabolisers (PMs) of nortriptyline are likely to require approximately 50 mg daily, whereas ultrarapid metabolisers (UMs) will require 150 mg. It has recently been shown that 18% of Swedes have an ultra-rapid CYP450 2C19 isoform (12) which may cause treatment failure with proton pump inhibitors and antidepressants. CYP2D6 substrates are all lipophilic bases with a protonable nitrogen atom, that includes all TCAs (13-15). However, it may be noted that there appears to be variation in substrate specificity for allelic variants (16) which means some drugs that are 2D6 substrates may be affected differently to others. Approximately 7% of the northern European population are deficient in the active form of this enzyme and are therefore poor metabolisers and develop high blood levels on standard doses, whereas 2% of that population (as you get less far from Out of Africa the % goes up, to 30+% in Ethiopia) have a gene multiplication making them ultrarapid metabolisers (UMs), and are therefore likely to have inadequate levels on standard doses. Look on David Flockharts website to give you an idea of the number of drugs affected by this http://medicine.iupui.edu/flockhart/clinlist.htm. Some narcotic analgesics, the codeine analogues and tramadol, are pro-drugs and are converted into a more active form by CYP450 2D6. That means many patients experience poor or no analgesic effect (17), and that fluoxetine and paroxetine inhibit the analgesic effect of these drugs. They will also, of course, push the level of all tricyclics higher, and possibly to toxic levels.
The above information highlights how therapeutic drug monitoring has been under utilized, remember my little rant above (18-20). It is essential to comprehend that those who are natural poor metabolisers or ultrarapid metabolisers (UMs) require the same care and monitoring for lack of effect, side effects and toxic effects as do those experiencing drug-drug interactions.
The table below summarises the most recent data concerning interactions. Although there are still significant gaps in our knowledge there is enough information to indicate fairly clearly which are the safe, or unsafe, combinations. You do not have to be a rocket scientist to see that fluvoxamine and fluoxetine stand out as very problematic drugs. In the case of fluvoxamine this is even more striking, because it is the only SSRI that various meta-analysis studies agree has the worst side effect profile of all the new drugs. It is difficult to conceive of a clinical situation in which fluvoxamine, or fluoxetine, would represent the best informed choice of treatment. Even in patients who are not taking any other drugs just 10 mg, that is one tenth of the minimum dose, produces significant reduction of the ability to metabolise caffeine, via CYP450 1A2, and makes everyone very sensitive to caffeine adverse effects. That may increase the risk of cardiac arrhythmias and possibly even spontaneous abortion (21-27).
Combinations that are likely to be safe with average doses of drugs are those involving sertraline (and possibly (es)citalopram) with an appropriate TCA, i.e. nortriptyline. If sertraline were mixed with dothiepin it is unlikely that significant problems would occur, but (es)citalopram, especially in higher doses may be risky. In my opinion nortriptyline has such clear pharmacological advantages over all the other TCAs that it represents the logical first choice in almost all clinical situations; amitriptyline can be used, but care is required with dosage and interactions, because of its ability to inhibit 2C19 and its inherently greater toxicity. Its 2C19 inhibition would block breakdown of some SSRIs, including sertraline, and in turn raise their levels, possibly pushing them to the point at which they themselves become significant inhibitors of 2D6. Doxepin should be regarded as a sedative, not an antidepressant, because its anti-histaminic properties are two orders of magnitude more potent than its noradrenaline reuptake inhibitor ability. At doses of 5 to 25 milligrams at night it is a potent and specific sedative with a wide margin of safety. Citalopram (and escitalopram) is a slightly more potent inhibitor of 2D6 than sertraline (28), and if used in doses above 20 milligrams per day this property becomes increasingly relevant. Although it is possible to use it, this can only be done with due recognition of this and other difficulties and limitations. There are several other general medical drugs that are significant inhibitors of 2C19. If one of these was added to treatment even the usually safe combination of sertraline plus nortriptyline might thereby be precipitated into producing a significantly increased side effects, and even problems.
Nortriptyline is the least problematic TCA of all, being only a weak 2D6 inhibitor. The tertiary amine TCAs, imipramine, clomipramine, amitriptyline, dothiepin are more potent, and significantly inhibit 2C19. Nortriptylines safety in overdose is good (5.5 deaths per million scripts, better than both venlafaxine at 13 and dothiepin at 53 (29,30)). These figures illustrate that the generalisation about TCAs being more toxic is inaccurate. Venlafaxine has a short half-life (and is unsatisfactory, except in slow release formulation, which has cost and other disadvantages) and has an especially troublesome withdrawal syndrome. Venlafaxines claim to be a dual-action antidepressant is founded on weak scientific evidence and its noradrenergic potency in humans is much less than nortriptyline; and there is no good replicated evidence that it is more effective than any other drug. Dothiepin is 10 times more likely to cause death in overdose compared to nortriptyline. A critical evaluation of dothiepins place in current therapy would put it in the same category as fluvoxamine, i.e. it is difficult to see a clinical situation in which it represents the optimal choice.
The table makes it easy to see that the combinations with the greatest potential dangers would probably be dothiepin with fluvoxamine or fluoxetine. Even small doses of these combinations would be predicted to result in major side effects. Average doses would be quite likely to precipitate significant toxicity, which in the case of dothiepin might involve life threatening cardiac arrhythmias. The predictable consequences of those combinations, and their potentially fatal nature, makes it difficult to argue against a blanket prohibition of their use. It places other health professionals in a less comfortable position in relation to knowing confidently what the best course of action is when confronted with such problematical combinations. Ultimately the answer to this question must be good communication between health professionals. In doubtful cases you may seek advice and 2nd opinion from the drug information services. However, incomplete knowledge and inconsistent, often out of date, information in supposedly authoritative texts can make it difficult to be sure of the best course of action.
Web sites recommended are:–
Human Cytochrome P450 (CYP) Allele Nomenclature Committee: http://www.cypalleles.ki.se/
Psychoactive Drug Screening Program (PDSP): http://pdsp.med.unc.edu/pdsp.php
CYP450 interactions http://medicine.iupui.edu/flockhart/clinlist.htm
Table: CYP450 enzyme inhibition of antidepressant drugs at usual therapeutic dose
|Drug||Cytochrome P450 enzyme inhibition|
- + = measurable, unlikely to be a clinically significant effect;
- ++ = clinically significant effects, possibly serious with other drugs with narrow safety margins;
- +++ = large, often clinically significant effect, serious interactions likely with susceptible drugs.
- +? Means no data is know to exist for that specific drug but its relationship to similar drugs with known effects suggests interactions are probable.
Data from references:–(11, 13, 28, 31-40)
*(41) . ** No known data
1. Anderson, I.M., Selective serotonin reuptake inhibitors versus tricyclic antidepressants: a meta- analysis of efficacy and tolerability. Journal of Affective Disorders, 2000. 58: p. 19-36.
2. McManus, P., et al., Co-prescribing of SSRIs and TCAs in Australia: how often does it occur and who is doing it? British Journal of Clinical Pharmacology, 2001. 51(1): p. 93-8.
3. Ellis, P., Australian and New Zealand clinical practice guidelines for the treatment of depression. Australian and New Zealand Journal of Psychiatry, 2004. 38(6): p. 389-407.
4. Preskorn, S.H., Detrimental Antidepressant Drug-Drug Interactions: Are they Clinically Relevant? Neuropsychopharmacology, 2006.
5. Gillman, P.K., Fluoxetine (Prozac). Medical Journal of Australia, 1993. 159(7): p. 492.
6. Medawar, P.S., Victims of Psychiatry. Vol. New York Review of Books Jan 23 17. 1975.
7. Kelly, W.N., Potential risks and prevention, Part 4: Reports of significant adverse drug events. Am J Health Syst Pharm, 2001. 58(15): p. 1406-12.
8. Preskorn, S.H., How drug-drug interactions can impact managed care. Am J Manag Care, 2004. 10(6 Suppl): p. S186-98.
9. Gillman, P.K., Drug interactions and fluoxetine: a commentary from a clinician’s perspective. Expert Opinion on Drug Safety, 2005. 4(6): p. 965-8.
10. Gillman, P.K., A review of serotonin toxicity data: implications for the mechanisms of antidepressant drug action. Biological Psychiatry, 2006. 59: p. 1046-51.
11. Preskorn, S. and D. Flockhart, 2006 Guide to Psychiatric Drug Interactions. Primary Psychiatry, 2006. 13: p. 35-64.
12. Sim, S.C., et al., A common novel CYP2C19 gene variant causes ultrarapid drug metabolism relevant for the drug response to proton pump inhibitors and antidepressants.Clinical Pharmacology and Therapeutics, 2006. 79(1): p. 103-13.
13. Ingelman-Sundberg, M., Genetic polymorphisms of cytochrome P450 2D6 (CYP2D6): clinical consequences, evolutionary aspects and functional diversity. The Pharmacogenomics Journal, 2005. 5(1): p. 6-13.
14. Johansson, I., et al., Inherited amplification of an active gene in the cytochrome P450 CYP2D locus as a cause of ultrarapid metabolism of debrisoquine. Proceedings of the National Academy of Sciences of the United States of America, 1993. 90(24): p. 11825-9.
15. Bertilsson, L., et al., Molecular genetics of CYP2D6: clinical relevance with focus on psychotropic drugs. British Journal of Clinical Pharmacology, 2002. 53(2): p. 111-22.
16. Bogni, A., et al., Substrate specific metabolism by polymorphic cytochrome P450 2D6 alleles. Toxicology In Vitro, 2005. 19(5): p. 621-9.
17. Tirkkonen, T. and K. Laine, Drug interactions with the potential to prevent prodrug activation as a common source of irrational prescribing in hospital inpatients. Clin Pharmacol Ther, 2004. 76(6): p. 639-47.
18. Mitchell, P.B., Therapeutic drug monitoring of non-tricyclic antidepressant drugs.Clinical Chemistry and Laboratory Medicine, 2004. 42(11): p. 1212-8.
19. Eap, C.B., E.J. Sirot, and P. Baumann, Therapeutic monitoring of antidepressants in the era of pharmacogenetics studies. Therapeutic Drug Monitoring, 2004. 26(2): p. 152-5.
20. Baumann, P., et al., The AGNP-TDM expert group consensus guidelines: therapeutic drug monitoring in psychiatry. Pharmacopsychiatry, 2004. 37(6): p. 243-65.
21. Bech, B.H., et al., Coffee and fetal death: a cohort study with prospective data. Am J Epidemiol, 2005. 162(10): p. 983-90.
22. Higdon, J.V. and B. Frei, Coffee and health: a review of recent human research. Crit Rev Food Sci Nutr, 2006. 46(2): p. 101-23.
23. Signorello, L.B., et al., Caffeine metabolism and the risk of spontaneous abortion of normal karyotype fetuses. Obstet Gynecol, 2001. 98(6): p. 1059-66.
24. Carrillo, J.A. and J. Benitez, Clinically significant pharmacokinetic interactions between dietary caffeine and medications. Clin Pharmacokinet, 2000. 39(2): p. 127-53.
25. Cannon, M.E., C.T. Cooke, and J.S. McCarthy, Caffeine-induced cardiac arrhythmia: an unrecognised danger of healthfood products. Med J Aust, 2001. 174(10): p. 520-1.
26. Katan, M.B. and E. Schouten, Caffeine and arrhythmia. Am J Clin Nutr, 2005. 81(3): p. 539-40.
27. Frost, L. and P. Vestergaard, Caffeine and risk of atrial fibrillation or flutter: the Danish Diet, Cancer, and Health Study. Am J Clin Nutr, 2005. 81(3): p. 578-82.
28. Preskorn, S.H., et al., The effect of duloxetine, escitalopram, and sertraline on CYP 2D6 function. Clinical Pharmacology and Therapeutics, 2006. 79: p. 52.
29. Buckley, N., Fatal toxicity of serotoninergic and other antidepressant drugs: analysis of United Kingdom mortality data. British Medical Journal, 2002. 325: p. 1332-1333.
30. Whyte, I.M., A.H. Dawson, and N.A. Buckley, Relative toxicity of venlafaxine and selective serotonin reuptake inhibitors in overdose compared to tricyclic antidepressants.Quarterly Journal of Medicine, 2003. 96(5): p. 369-74.
31. Preskorn, S.H., Reproducibility of the in vivo effect of the selective serotonin reuptake inhibitors on the in vivo function of cytochrome P450 2D6: an update (part II). Journal of Psychiatric Practice, 2003. 9(3): p. 228-36.
32. Preskorn, S.H., Reproducibility of the in vivo effect of the selective serotonin reuptake inhibitors on the in vivo function of cytochrome P450 2D6: an update (part I). Journal of Psychiatric Practice, 2003. 9(2): p. 150-8.
33. Shin, J.G., et al., Inhibitory effects of tricyclic antidepressants (TCAs) on human cytochrome P450 enzymes in vitro: mechanism of drug interaction between TCAs and phenytoin. Drug Metabolism and Disposition: The Biological Fate of Chemicals, 2002. 30(10): p. 1102-7.
34. Danie, W.A., et al., Effects of antidepressant drugs on the activity of cytochrome P-450 measured by caffeine oxidation in rat liver microsomes. Polish Journal of Pharmacology, 2001. 53(4): p. 351-7.
35. Szewczuk-Boguslawska, M., et al., Doxepin inhibits CYP2D6 activity in vivo. Polish Journal of Pharmacology, 2004. 56(4): p. 491-4.
36. Skinner, M.H., et al., Duloxetine is both an inhibitor and a substrate of cytochrome P4502D6 in healthy volunteers. Clinical Pharmacology and Therapeutics, 2003. 73(3): p. 170-7.
37. Vandel, P., et al., Clomipramine, fluoxetine and CYP2D6 metabolic capacity in depressed patients. Hum Psychopharmacol, 2004. 19(5): p. 293-8.
38. von Moltke, L.L., et al., Inhibition of alprazolam and desipramine hydroxylation in vitro by paroxetine and fluvoxamine: comparison with other selective serotonin reuptake inhibitor antidepressants. Journal of Clinical Psychopharmacology, 1995. 15(2): p. 125-31.
39. Albers, L.J., et al., Effect of venlafaxine on imipramine metabolism. Psychiatry Research, 2000. 96(3): p. 235-43.
40. Brosen, K., Some aspects of genetic polymorphism in the biotransformation of antidepressants. Therapie, 2004. 59(1): p. 5-12.
41. Salsali, M., A. Holt, and G.B. Baker, Inhibitory effects of the monoamine oxidase inhibitor tranylcypromine on the cytochrome P450 enzymes CYP2C19, CYP2C9, and CYP2D6. Cell Mol Neurobiol, 2004. 24(1): p. 63-76.