Psychopharmacological Neuroprotection in Neurodegenerative Disease: Assessing the Preclinical Data

A National Library of Medicine PubMed search was conducted October 1, 2007. Specific National Library of Medicine search terms utilized were: alpha-synuclein, beta-amyloid, tau, ubiquitin, proteasome, mitochondrial viability, mitochondria, mitochondrial transition pore, cytochrome c release, leukocyte viability, and apoptosis. We examined the following drugs, which have been routinely applied as first-line symptomatic treatments for neuropsychiatric disorders arising in the context of neurodegenerative disease: apathy treatments (pramipexole, ropinirole, amantadine), antipsychotics (haloperidol, fluphenazine, trifluoperazine, thiothixene, chlorpromazine, thioridazine, risperidone, olanzapine, quetiapine, ziprasidone, aripiprazole, clozapine), mood stabilizers (lithium, carbamazepine, oxcarbazepine, valproate), antidepressants (amitriptyline, imipramine, nortriptyline, desipramine, clomipramine, trimipramine, doxepin, protriptyline, maprotiline, bupropion, fluoxetine, sertraline, fluvoxamine, paroxetine, citalopram, s-citalopram, trazodone, nefazodone, venlafaxine, duloxetine, mirtazapine), anxiolytics/hypnotics (buspirone, diazepam, chlordiazepoxide, flurazepam, temazepam, chlorazepate, clonazepam, lorazepam, oxazepam, alprazolam, zaleplon, zolpidem, zopiclone, s-zopiclone), antihistamines (cyproheptadine, hydroxyzine), anticholinergics (benztropine, trihexyphenidyl), modafinil, melatonin, and ramelteon. In light of the focus of the search on common first-line psychotropics used in clinical practice to treat the behavioral and neuropsychiatric manifestations of neurodegenerative disease, certain drugs were not included in the search. For example, although certain drugs have psychotropic properties, we did not include cognitive enhancers (cholinesterase inhibitors, memantine, nootropics), mono-amine oxidase inhibitors (e.g., selegeline, rasagiline), or other anticonvulsants (e.g., lamotrigine) because these drugs do not meet the criterion of “first-line psychotropics.” The search strategy joined terms of interest with specific drugs by means of an “AND” operator. All peer-reviewed articles in the database with English abstracts on or before October revealed by the search were considered; the current review (part one) was limited to articles within the database; part two of this report employed the findings of part one extended by review of bibliographies and other sources including additional National Library of Medicine searches of the literature.

The preclinical investigation of the impact of psychotropic drugs on molecular processes pertinent to neuroprotection varied considerably. For example, regarding dopamine agonists, we identified one paper addressing impact on αSyn, two on Aβ, one on mitochondrial function, eight on the permeability transition pore, and eight on apoptosis. No papers were identified addressing the impact of these drugs on tau, proteasomes, or cell viability. By comparison, with regard to antipsychotics, we identified six papers addressing effects on Aβ, five on tau, four on proteasomes, 28 on mitochondrion, 25 on permeability transition pore, 11 on cell viability, and 45 on apoptosis, yet no papers discussed the impact of these agents on αSyn. Only 10 total papers were identified addressing the effects of all these psychotropics on ubiquitin—one indication of the weakness of the literature in certain areas. In addition, there was considerable variability in the laboratory approaches, models, and assays utilized to examine the impact on a given molecular process. For instance, studies of mitochondrial effects used mouse, rat, or human brain cell cultures, mouse or human heart, liver or endothelial cells, and normal or neoplastic leukocytes. These studies variously assessed oxygen uptake, Complex I, II, IV, or V activity, ATP production, succinate production or succinate dehydrogenase activity, redox reaction velocity, reactive oxygen species production, and/or morphological changes on electron microscopy. Even within the papers that focused on human brain cells, different models used a variety of neuron types including those from brainstem, basal ganglia, cerebellum, and several regions of the cortex. We organized and summarized the available data making no assumptions about relative predictive translational neuroprotective merits of different models and tissues, which are not known at present (see discussion).

The most important detailed findings for each drug are briefly summarized in Table 1, Table 2, Table 3, and Table 4 (located online at http://neuro.psychiatryonline.org/cgi/content/full/22/1/8/DCI). The recently discovered TDP-43 was also considered while this project was underway, but no relevant articles were evident for this protein.

It is evident from the above that there is significant variation in degree of investigation, cell lines studied, and methodological approaches. Other limitations include the varying use of neural tissues, variance in the neuronal types studied, use of neuroblastoma lines instead of neurons, study of immature or poorly differentiated cells that may be more prone to apoptosis than more mature cells, and the infrequent characterization of effects on αSyn, tau, and Aβ. Such deficiencies in the data significantly confound the ability to draw definitive conclusions. In particular, the deficiencies in the data raise the question as to the most valid, clinically relevant, and appropriate standards of evidence to apply in determining which preclinical findings will predictably translate into clinical neuroprotection in patients with neurodegenerative diseases.

A number of concerns impact the selection of an appropriate standard of evidence. First, there are no established general criteria for judging preclinical neuroprotective data across the diversity of neurodegenerative diseases. Second, unlike clinical evidence-based medicine (EBM) standards, there do not appear to be established uniform criteria for judging the diversity of preclinical findings. From an EBM perspective, the data considered here are even less compelling than Class II or IV 18 or Level C 19 clinical case reports since they generally do not pertain to findings in human patients. Third, there are considerable variabilities across the present preclinical findings with respect to intra- and extramodel replication, replications in neural tissue, the specific neural tissues studied, and the specific brain locus even when neurons are consistently studied. These are summarized in Table 5. Fourth, replications are still needed using the same physiological dose range, particularly because some have observed bell—shaped rather than sigmoid—shaped neuroprotective dose—response curves. 20, 21 Fifth, some drugs have mixed actions, simultaneously possessing some neuroprotective actions and other neurodegenerative actions. It is not yet clear whether the various actions should receive equal weight or whether one may trump others (for example, effects on apoptotic measures may be more determinative in importance than effects on more “upstream” processes such as mitochondrial potential or proteasomal function). Sixth, there is no gold-standard preclinical model but, instead, a diversity of models that each have their own select benefits and limitations. These and other factors likely contribute to the current disconnect between preclinical findings and neuroprotective clinical trial results.

Some criteria for considering neuroprotective candidate agents have been elaborated in Parkinson’s disease 22 and stroke. 23 In Parkinson’s disease, scientific rationale, penetration of the blood-brain barrier, safety and tolerability, and efficacy in relevant animal models of the disease or an indication of benefit in human clinical studies constitute criteria. 22 In the case of FDA-approved psychotropics reviewed here, which essentially meet most of these criteria (with the exception of systematic, consistent application in relevant neurodegenerative disease models), the question then becomes: how good is the available preclinical evidence of neuroprotection? Ravina et al. 22 noted that the most problematic issue in Parkinson’s disease was evaluating animal data given the many different models that were of uncertain value in predicting results in humans and noted further that a clinical trial would actually be needed to demonstrate the predictive validity of any preclinical model. Similarly, it is not possible to judge the quality of the present preclinical findings by the models used because the predictive validities of the models remain unclear. In stroke, 23 potentially successful drug candidates have been considered to be inferable from preclinical data by the following criteria: (a) adequately defined dose-response relations; (b) time window studies showing a benefit period; (c) adequate physiological monitoring in unbiased, replicated, randomized, blinded animal studies; (d) lesion volume and functional outcome measures determined acutely and at longer term followup; (e) demonstration in two animal species; (f) submission of findings to a peer-reviewed journal. However, even with these criteria, Gladstone et al. 24 have pointed out that translation of preclinical findings to clinical efficacy has been hampered by a lack of functional outcomes, long-term end points, permanent ischemia models, extended time windows, and selective white matter evaluation in preclinical models whereas clinical studies are plagued by insensitive outcome measures, lack of stroke subtype specificity, and inattention to the ischemic penumbra, among other concerns. Ford 25 has also pointed out that a number of compounds fulfilling these stroke neuroprotectant criteria have failed to afford translational clinical neuroprotection. Analogous concerns obtain for neurodegenerative disease preclinical models and clinical methods, particularly whether putative criteria will reliably predict translation to clinical neuroprotection. Additionally, a nearly endless array of clinical variables including gender, age, pharmacogenomics, medical history, coadministered drugs, and other factors may contribute to an inability to predict clinical neuroprotection despite preclinical success. Thus, predictive criteria remain in need of development.

Reflection upon these translational issues in regard to psychotropic neuroprotection in neurodegenerative diseases first suggests the need for replication within and between specific preclinical models in specific neurons at specific loci to elucidate physiological dose-response relations that should then themselves be replicated as a first step. Additionally, other issues seem relevant to the problem of determining which candidate drugs may be most likely to effect clinical neuroprotection. We suggest preliminary neuroprotective drug selection criteria for assessing the likelihood of translational clinical neuroprotection in neurodegenerative diseases (Table 6). These criteria, including preclinical (at least two replicated neuroprotective actions at physiological doses in an established neuroprotective model, neural tissue, and disease-specific animal model in excess of the number of known neurodegenerative actions) and clinical (delayed progression on clinical markers and unexpected benign disease course not accounted for by symptomatic properties) criteria, can be evaluated over time and modified as future data indicate. Given the lack of information regarding the utility of specific preclinical paradigms in predicting clinical neuroprotective effects, it is premature to rank or weight these criteria. Rather, recent concerns 26 notwithstanding and until a better study methodology is developed, we suspect that the greater the number of criteria met by a candidate drug, the greater the likelihood of demonstrating translational clinical neuroprotective efficacy in a randomized, double-blind, placebo-controlled, delayed-start or randomized-withdrawal clinical trial. 27 Such trials are needed because agents deemed promising based upon preclinical data often fail to demonstrate neuroprotection in clinical trials for reasons identified in the above paragraph. At present, preclinical demonstration of replicable neuroprotective effects in neural tissues at clinically-relevant doses does not assure a positive result in a clinical trial, nor does the absence of such evidence necessarily exclude clinical neuroprotective benefits. Until such clinical findings obtain, it is impossible to identify preclinical determinants predictive of translational clinical success and ascertain whether patients are actually being helped or harmed in a neuroprotective sense by the use of these drugs.

Beyond the methodological concerns expressed above, a practical assessment of these preclinical findings is still possible. Given the relative infancy of this field of research, the present state of the literature, the limitations of the data described above, and our current ignorance of preclinical evidence predictive of successful clinical translation, there is the very real possibility of prematurely disregarding findings that may ultimately prove to be of clinical significance with further research (a “type II” error) by applying an overly stringent standard of evidence. It seems that, at the present time, the proper approach is to instead look at the preponderance of the available findings and attempt some generalizations that constitute general impressions to be tested in future research, similar to the process of developing and refining clinical diagnostic criteria. Accordingly, the following observations are drawn from looking at all of the studies, without any exclusions, except where there are clearly contradictory data. As noted, many of the findings have not yet been independently replicated in the same model despite apparent replication in a different model (Table 5). Until the state of the literature develops to the point where independent replications in the same model are routinely observed, appropriate assessment criteria must be very liberal, resulting in conclusions that can only be viewed as preliminary. Adopting this approach with its attending caveats, some preliminary observations can be gleaned from the data. Below, we first consider drugs with respect to their neuroprotective potentials, distinguishing drugs meriting further study from those that have limitations dissuading further investigation and those for which too little data are available to form any conclusions. (We also summarize neuroprotective effects by drug class in Appendix 1 and drugs by neuroprotective actions in Appendix 2 [located online at http://neuro.psychiatryonline.org/cgi/content/full/22/1/8/DCI]; Part 2 of this report focuses on the broader neuroprotective aspects of selected psychopharmacological classes.) Next, we assess the general properties of the various classes of psychotropics. We then consider each investigated cellular function with regard to the drugs that influence them. Finally, we detail a research agenda for drugs of interest and consider the progress made in clinical neuroprotective trials thus far, recommending a next step in their development.

This report was presented in part at the 20th Annual Meeting of the American Neuropsychiatric Association, San Antonio, TX, February 18–22, 2009. The authors wish to thank Ms. Karen J. Hall for her assistance in manuscript preparation.

Part II will further explore heuristic clinical applications of psychopharmacological neuroprotection in neurodegenerative disease. Look for it in the Spring 2010 issue of The Journal of Neuropsychiatry and Clinical Neurosciences, available at http://neuro.psychiatryonline.org/.

Full text of all data tables and appendices referenced in this article, along with the complete 548 references, accompanies the online edition of this issue at http://neuro.psychiatryonline.org/cgi/content/full/22/1/8/DCI.