Neuroprotection in Parkinson Disease: Title and subTitle BreakMysteries, Myths, and Misconceptions

Parkinson disease (PD) is the second most common neurodegenerative disease after Alzheimer disease, having an incidence of approximately 1 in 200 and a lifetime risk of 1 in 45. Approximately 1 million people have PD in the United States, and the prevalence is predicted to increase dramatically in the coming decades due to the aging of the population.¹ Parkinson disease is responsible for significant morbidity and increased mortality among those with the disease and causes substantial social and economic hardship for patients, caregivers, and society.

The cardinal clinical features of PD are resting tremor, rigidity, and bradykinesia or motor slowing. Pathologically, the disease is characterized by a preferential loss of dopamine neurons in the substantia nigra pars compacta (SNc), with intracellular proteinaceous inclusions or Lewy bodies and a reduction in striatal dopamine. In the early stages of the disease, symptoms are relatively well controlled with levodopa and other dopaminergic agents. However, over time the majority of patients experience levodopa-related motor complications and the disease relentlessly progresses with the development of features such as freezing, falling, autonomic dysfunction, and dementia that do not adequately respond to dopamine replacement therapies.¹ Thus, despite advances in modern therapy, patients with PD continue to experience unacceptable disability. The main challenge facing those involved in the management of patients with PD is the development of a neuroprotective therapy that can be administered early in the course of the disease and slow, stop, or reverse disease progression.

Recent studies have drawn attention to the potential for certain drugs to provide neuroprotective benefits in PD. These results have attracted considerable attention and debate. At the heart of this controversy is the issue of determining whether the clinical and imaging end points that have been used in these trials do in fact measure disease progression. This review seeks to set this debate in context, addressing the scientific, clinical, and pragmatic dimensions of an increasingly important and complex area of disease management. Studies included in this review are based on an extensive search of PubMed, 1989 to November 2003, on the terms neuroprotection or disease modification and PD. Trials included were prospective, placebo-controlled, double-blind trials of putative neuroprotective therapies for PD.²

The development of a disease-modifying or neuroprotective agent for PD would be greatly facilitated by understanding the mechanism responsible for neuronal degeneration. The cause of neurodegeneration in PD likely is multifactorial in terms of both etiology and pathogenesis. Genetic factors are now known to cause PD in small numbers of patients with a familial form of the disorder. Mutations in 5 different genes (alpha-synuclein, parkin, UCH-L1, DJ-1, and Nurr-1) have been identified, and PD has been linked to an additional 5 different chromosomal loci.³ In contrast, twin studies suggest that while genetic factors may be important in young-onset patients, they do not play a primary role in the large majority of individuals who experience a sporadic form of the disorder most likely related to environmental factors.⁴

Indeed, current thinking favors the hypothesis that most sporadic cases are caused by a complex interplay between different genetic and environmental factors. Several biochemical factors appear to be involved in the pathogenetic cascade of events leading to cell dysfunction and death in PD, including free radicals, a mitochondrial complex I deficiency, excitotoxicity, and inflammation.⁵ Here too, it appears that the extent to which each of these factors contributes to the neuronal degeneration varies in an individual patient. It now appears that protein mishandling and aggregation contribute to the neurodegenerative process and may represent a common feature in the different etiopathogenetic forms of PD.⁶ Finally, increasing evidence suggests that cell death in PD occurs by way of a signal-mediated apoptotic process, although the precise signals that are involved have not yet been completely defined.⁷ Thus, laboratory studies have already delivered an attractive array of molecular targets and candidate drugs that might be neuroprotective in PD (Figure 1).

Antioxidants were the first drugs to be studied in an attempt to retard the progress of PD. The Deprenyl and Tocopherol Antioxidative Therapy of Parkinsonism (DATATOP) study was a prospective double-blind, placebo-controlled trial that evaluated the antioxidant vitamin E in a total daily dose of 2000 IU and the monoamine oxidase type B (MAO-B) inhibitor deprenyl 5 mg twice daily as putative neuroprotective therapies.⁸ The primary end point was time until untreated patients with PD required levodopa. No beneficial effect of vitamin E was detected, although it is possible that there was poor brain penetration or inadequate dosing. In contrast, deprenyl significantly delayed the need for levodopa compared with placebo, consistent with slowing of disease progression. However, deprenyl was also found to exert a mild symptomatic effect that confounded interpretation of the study. Thus, it was not possible to determine if the delay in the need for levodopa was because the drug slowed neuronal degeneration or because symptomatic effects masked ongoing disease progression. To try to avoid this confound, deprenyl was compared with placebo using the change in motor score between an untreated baseline visit and a final visit at 14 months performed after 12 months of treatment with study drug and 2 months of study drug withdrawal as the primary end point.⁹ Patients treated with deprenyl had less deterioration from baseline than patients treated with placebo, again suggesting that deprenyl might be neuroprotective. Here too, however, the study may have been confounded by the potential of deprenyl to have long-lasting symptomatic effects.

More recently, coenzyme Q₁₀ was studied as a putative neuroprotective agent in PD. It is an intrinsic component of the mitochondrial respiratory chain that functions both as an enhancer of adenosine triphosphate (ATP) production and an antioxidant. As there is both a complex I defect and oxidative damage in the SNc in PD, this compound was a logical candidate for study. A double-blind, placebo-controlled pilot study demonstrated that high doses of coenzyme Q₁₀ (1200 mg/d) were associated with a reduced rate of deterioration in motor function from baseline over the 16-month course of this trial.¹⁰ However, activity of daily living scores improved significantly following introduction of the drug, raising the possibility that an unanticipated symptomatic effect might also have confounded interpretation of this study.

Thus, the clinical end points that have been used to evaluate the effect of a putative neuroprotective drug on disease progression in PD are readily confounded by any symptomatic benefit of the study intervention. This has been a problem with deprenyl and coenzyme Q₁₀, which were not anticipated to have a symptomatic effect in PD, and obviously is an even greater problem when testing possible neuroprotective agents that are known to have antiparkinsonian effects. This issue has led to a search for a nonclinical, objective, surrogate marker of disease progression that can be used as an end point in studies of possible disease-modifying therapies.

Neuroimaging can provide a marker of the functional integrity of the nigrostriatal system by measuring either striatal uptake of fluorine 18 fluorodopa with positron emission tomography (FD-PET) or 2β-carbomethoxy-3β-[4-iodophenyl]tropane (β-CIT) uptake with single-photon emission computed tomography (β-CIT SPECT) (Figure 2). Neuroimaging with FD-PET assesses the capacity of dopaminergic neurons to decarboxylate and store levodopa/dopamine, while β-CIT SPECT reflects the density of dopamine transporters (DATs) on presynaptic dopamine terminals. Both of these techniques show a characteristic pattern of asymmetric signal loss primarily in the posterior putamen in patients with PD (Figure 2). Longitudinal studies have shown that both of these markers decline linearly with disease progression in patients with PD, and cross-sectional studies demonstrate a correlation with bradykinesia scores in advanced cases in which compensatory changes are less likely to occur.^{11 - 16} Based on these considerations, the rate of decline in striatal fluorodopa uptake on PET and β-CIT on SPECT have been used as primary end points for testing the putative neuroprotective effect of dopamine agonists.

Dopamine agonists have been used for many years in the symptomatic treatment of PD. They are well tolerated and effective antiparkinsonian agents and, in particular, are associated with a significant reduction in the rate of development of motor complications in comparison with levodopa.^{17 - 18} They also exhibit properties that have attracted the attention of those attempting to develop neuroprotective therapies, in that they are antioxidants, decrease the turnover of levodopa and so reduce free radicals generated by its metabolism, may diminish excitotoxicity due to overactivity of the subthalamic nucleus that occurs in PD, and demonstrate antiapoptotic effects in a variety of cell models.¹⁹ In the laboratory, dopamine agonists protect dopaminergic and nondopaminergic neurons from a variety of toxins in both in vitro and in vivo models of parkinsonism.^{19 - 23}

Based on these findings, 2 prospective double-blind clinical trials have been performed testing the capacity of dopamine agonists to modify the rate of disease progression in newly diagnosed patients with PD using imaging markers as primary end points. The first of these studies (CALM-PD-CIT) randomized patients with newly diagnosed PD to initial treatment with either the dopamine agonist pramipexole or levodopa.²⁴ Striatal β-CIT uptake on SPECT to determine DAT density was performed at baseline and at various times during the 4-year trial. Patients in either group who required additional symptomatic therapy could be supplemented with open-label levodopa. Patients randomized to receive pramipexole had a significant 36% reduction in the rate of decline of β-CIT uptake compared with those initiated with levodopa therapy (Figure 3). In a second study (Requip as Early Therapy versus L-dopa [REAL-PET]), patients with untreated PD were randomized to begin therapy with the dopamine agonist ropinirole or levodopa.²⁵ In this study, patients randomized to begin treatment with the agonist had a significant 35% reduction in the rate of decline in striatal FD uptake on PET compared with those randomized to start therapy with levodopa (Figure 3). Interestingly, there was no corresponding clinical benefit in favor of the dopamine agonists to go along with these imaging findings in either study.

These studies have generated considerable interest and debate, if not controversy. Both demonstrate that dopamine agonists are associated with a significant delay in the rate of decline of a surrogate imaging biomarker of nigrostriatal function (Figure 3). One interpretation of these findings is that these 2 dopamine agonists slow the rate of cell loss in the SNc of patients with PD, consistent with evidence of their neuroprotection effect in the laboratory. However, neither showed a corresponding clinical benefit, although it can be argued that the time course of the trials may have been too short to permit such an effect to be detected in the context of viable compensatory mechanisms and powerful symptomatic agents, and this will only become apparent with longer follow-up.

Another interpretation of these studies that could account for the different rates of decline in patients treated with levodopa and dopamine agonists is that levodopa is toxic to SNc neurons. There is a concern that levodopa might be toxic to dopamine neurons as it undergoes oxidative metabolism and has the potential to generate cytotoxic free radicals.²⁶ In the laboratory, levodopa can be toxic to cultured dopamine neurons, but there is no convincing evidence that levodopa is toxic in vivo and especially in patients with PD.²⁷ The recently completed Early versus Late L-dopa in Parkinson Disease (ELLDOPA) study²⁸ investigated the effect of levodopa vs placebo on disease progression. The primary end point was the change in motor score between untreated baseline and a final visit, performed after 9 months of study drug treatment and 2 weeks of drug withdrawal. At the completion of the trial, patients who had been randomized to receive levodopa had less deterioration from baseline than did those receiving placebo. While these results possibly could relate to a benefit of levodopa that persists for weeks after the drug has been withdrawn, these results do not suggest that levodopa is toxic. However, as part of this study, a subgroup of patients underwent β-CIT SPECT scans at baseline and at 9 months. Patients treated with levodopa had a greater rate of decline in this imaging biomarker than did those in the placebo group, consistent with levodopa having a toxic effect. Thus, these apparently conflicting results do not permit a clear determination of whether levodopa is toxic and might have adversely affected dopamine neurons in the CALM-PD-CIT or REAL-PET studies.

Finally, it has been proposed that the differences between the effects of levodopa and dopamine agonists seen in the CALM-PD-CIT and REAL-PET studies are not related to any direct effect of the drugs on dopamine neuron survival or degeneration, but rather to a pharmacological difference in the capacity of these drugs to regulate DAT or FD metabolism.^{29 - 31} However, studies testing the effects of levodopa and dopamine agonists on DAT and FD metabolism are conflicting.

Dopamine transporters, which remove dopamine from the synapse, might be expected to be downregulated in PD, to augment dopamine availability.^{32 - 34} However, this effect should be similar with both agonist and levodopa treatment. Dopamine agonists have been reported to upregulate DAT in in vitro cell models^{35 - 36} but to have no effect on DAT expression in short-term studies in rodents.^{37 - 38} Similarly, short-term levodopa treatment has been reported to produce no change,^{39 - 41} upregulation,⁴² and downregulation³⁹ in DAT expression. In another study, administering levodopa to rodents for 6 months had no effect on DAT expression in either controls or animals with severe 6-hydroxydopamine–induced lesions, but expression was increased in less severely lesioned animals⁴³ and in 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP)-lesioned monkeys that had been treated with levodopa for 9 weeks.⁴⁴ In summary, "the results of carefully controlled animal studies have yet to provide any consensus on the direction of the change in striatal DAT levels caused by drugs that either enhance (L-dopa, dopamine agonists, DAT blockers) or decrease (dopamine-depleting agents) dopaminergic function."⁴⁵

The data in humans are similarly inconsistent. In the CALM-PD-CIT trial, a small group of patients were scanned 10 weeks after initiation of treatment to determine if there were any short-term effects on DAT expression. No change was detected with either levodopa or pramipexole. Similarly, levodopa had no effect on DAT in 2 other small, short-term studies in patients with PD.^{46 - 47} In another short-term but larger study, both levodopa and pramipexole reduced DAT binding in patients with PD, although the effect was greater with levodopa.⁴⁵ Taken together these results do not show a consistent effect of either dopamine agonists or levodopa on DAT expression in vitro or in vivo. However, it must be appreciated that these studies were all relatively small scale and performed over relatively short periods of time.

The data on the potential for dopamine agonists and levodopa to modulate fluorodopa metabolism are similarly limited. In animal studies, both levodopa and dopamine agonists downregulate striatal dopa decarboxylase activity.^{48 - 49} In patients with PD, acute administration of either a dopamine agonist or levodopa reduces L-[¹¹C]dopa uptake on PET,^{50 - 51} probably through activating presynaptic autoreceptors leading to decreased dopamine synthesis and release^{52 - 54} and downregulation of dopa-decarboxylase.^{48 ,55 - 56} Thus, the data available indicate that both dopamine agonists and levodopa likely have the same effect on PET imaging.

The changes described above suggest that the differences observed in the imaging studies are not due to excess downregulation of DAT or decarboxylase by levodopa compared with dopamine agonists, although that possibility cannot be excluded. It should be noted, however, that identical results were obtained in the REAL-PET and CALM-PD-CIT studies despite the use of different agonists, different tracers, and different imaging techniques that tend to regulate the nigrostriatal system in opposite directions (ie, dopamine depletion is associated with upregulation of decarboxylase and downregulation of DAT). Further, in each of the REAL-PET, CALM-PD-CIT, and ELLDOPA studies, several patients had normal studies at baseline, presumably because of misdiagnosis, and in none of these patients did chronic treatment with either levodopa or an agonist result in any change in striatal FD uptake on PET or β-CIT uptake on SPECT.

In summary, both clinical and imaging end points that have been used to try to detect a putative neuroprotective effect in PD are potentially confounded. Additional trial designs have been proposed to try to resolve these issues.⁵⁷ One involves a randomized washout design that aims to detect a difference from baseline between study drug and placebo that persists for a prolonged period of time after washout and therefore could be deemed to represent a neuroprotective effect. Unfortunately, this design does not lend itself to studies of PD for which prolonged washout usually is not feasible.

More promising is the randomized start design that has been recently used in trials of the MAO-B inhibitor rasagiline.⁵⁸ In this study design, patients are randomized to receive study drug or placebo for a fixed time interval after which patients in the placebo group are also initiated on study drug. If an initial difference between the 2 groups endures even after patients in the placebo group are started on study drug, this could be interpreted to represent a neuroprotective benefit, since the placebo group would be expected to catch up if the effect was entirely symptomatic. This approach would also allow for patients in both treatment groups to receive imaging studies at baseline when both are untreated and at a later time point when both are receiving the study drug. This approach would diminish the likelihood that differences between groups were due to regulatory effects.
 Yet another approach is to compare the long-term outcomes with respect to nondopaminergic features such as posture instability and dementia in patients randomized to receive placebo or study drug, as it is unlikely that benefits in these features could be attributed to a symptomatic effect. Finally, it may be necessary to use multiple primary end points with appropriate adjustment for multiple comparisons and to require positive results in a variety of parameters to conclude that a given drug is associated with a neuroprotective effect.

The dilemma now is how physicians who treat patients with PD can place this information in context. Neuroprotective treatment is essential in order to limit the disability experienced by millions of patients with PD. Laboratory studies have provided numerous candidate neuroprotective drugs, but clinical end points that have been used to date are readily confounded by any symptomatic effect of the study intervention and thus do not provide an unequivocal measure of disease progression that can be used to determine if a drug has a neuroprotective effect. Surrogate neuroimaging markers have been used to try to circumvent this problem, and 2 recent trials have reported that patients randomized to receive treatment with a dopamine agonist had a reduced rate of decline in these measures of nigrostriatal function compared with levodopa. Laboratory studies suggest that these results could be due to a protective effect of dopamine agonists or a toxic effect of levodopa, but it is not possible to differentiate these as neither study included a placebo control. It has also been suggested that the study results may be due to pharmacological differences in the capacity of these agents to regulate components of the nigrostriatal system.

Critical review of the available in vitro, animal, and human data, however, are conflicting and inconclusive. Accordingly, we believe that the fairest position at this time is that there is insufficient information to draw the conclusion that the results of the pramipexole and ropinirole trials have been caused by spurious pharmacological effects, but neither can one say with certainty that the agonists are protective or that levodopa is toxic. Nevertheless, the combination of in vitro and in vivo laboratory evidence demonstrating a neuroprotective effect of dopamine agonists, together with the results of the CALM-PD-CIT and REAL-PET studies, make a compelling story that should stimulate further research.

For now, there is no clear answer regarding neuroprotection for the patient with early PD. The decision to introduce a putative neuroprotective therapy remains a matter of judgment and the personal philosophy of the patient and the treating physician. However, patients and physicians alike should draw considerable encouragement from the recent advances in understanding of the causes of PD and from the ability to turn these into potential therapies that might prevent the progression of the disease. The most immediate challenge is to define a reliable method to detect the rate of disease progression that can be used to assess putative neuroprotective agents. The seminal studies described above are likely to be the first of many such studies to attempt to identify neuroprotective agents for the treatment of PD.