Argument 1: levodopa itself is primarily responsible for dyskinesias and motor fluctuations

A commonly shared opinion is that levodopa itself is primarily responsible for the development of motor fluctuations. This idea is based on the theory that oral levodopa administration, together with its short elimination half time, leads to unnatural ‘pulsatile’ dopaminergic stimulation of the dopamine receptors, unlike the more continuous physiological stimulation in healthy subjects.6 Moreover, the appearance of these motor fluctuations seems to be related to the absolute levodopa dose and the duration of levodopa administration.4 8

However, it is difficult to interpret the data that support this argument. Patients taking high doses of levodopa, or patients with a prolonged need for levodopa treatment, probably also have more severe Parkinson's disease. No study has yet addressed the best method by which levodopa should be administered to optimise efficacy and minimise motor complications.9 In uncontrolled follow-up studies, it is hardly possible to separate the effects of therapy from those of disease severity.

One interesting exception was the observation made in the 1980s in California. A group of young people developed severe, subacute parkinsonism following intoxication with the heroin analogue MPTP (1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine). These patients responded excellently to levodopa but they also developed motor fluctuations shortly after starting the drug.10 11 This suggests that disease severity at the time of levodopa initiation is one of the main causes.

This theory is also supported by Kostic et al who tried to separate the severity of Parkinson's disease from the adverse effects of levodopa.12 If levodopa was introduced in advanced Parkinson's disease, patients developed motor fluctuations quicker than if it had been introduced at an earlier stage of the disease. In a recent study, patients without any motor fluctuations were compared with patients with mild or severe motor fluctuations.13 Patients with a younger age of onset (and therefore a longer disease course at the time of examination) more frequently showed motor fluctuations. It was interesting to note that the mean age of death was almost the same in both groups. The four milestones of advanced disease (frequent falls, visual hallucinations, cognitive disability and need for residential care) also occurred at a similar time before death (5–7 years) in both groups.13 These data suggest that the appearance of fluctuations and disabling signs in the course of the disease is mainly determined by disease progression and not by levodopa therapy.

Argument 2: levodopa is toxic

There has been considerable debate about whether or not levodopa is neurotoxic. In vitro, toxic free radicals are formed with oxidation of dopamine.2 6 However, in daily practice there is no evidence that this translates to human disease, as reflected by the following observations:

• The introduction of levodopa coincided with a decrease in Parkinson's disease mortality14

• The Earlier versus Later L-DOPA (ELLDOPA) trial compared the effect of three different levodopa doses in previously untreated Parkinson's disease patients.15 After a treatment period of 55 weeks and a washout period of 1 month, there was no visible deterioration of the levodopa groups compared with the placebo group (in fact, if anything the results suggested the opposite).15 This study also included nuclear imaging of dopamine reuptake status and this did hint at a toxic levodopa effect. However, there is now growing consensus that such scans are very difficult to interpret in the face of symptomatic drug treatment.

• Early restoration of basal ganglia physiology by starting dopaminergic drugs may support compensatory neural mechanisms and thereby delay potentially irreversible modification of neural circuitries involved in the pathophysiology of Parkinson's disease.16 If shown to be true, such an effect should theoretically afford a lasting clinical benefit for patients treated early in the course of their disease. However, recent ‘early start’ trials (the Attenuation of Disease Progression with Azilect Given Once daily (ADAGIO) trial with rasagiline and the Pramipexole On Underlying Disease (PROUD) trial with pramipexole)17 18 had conflicting data that were difficult to interpret. The ADAGIO trial was able to show a persistent difference between early and delayed start of rasagiline at a dose of 1 mg daily but oddly enough not at 2 mg.17 The PROUD trial did not show any difference between early versus late start treatment groups.18 This means there may be drug class differences with respect to a possible early restoration effect. No comparable data are available on the early or delayed start of levodopa. Although interesting, this disease modification hypothesis remains unproven. At the very least, the studies showed no evidence whatsoever that early treatment was harmful although admittedly follow-up was relatively brief.

Argument 3: levodopa loses its effectiveness

There is a persistent misunderstanding that the effectiveness of levodopa declines or even disappears with time, but there is no scientific evidence to support this position. Although the disease will inevitably continue to progress, levodopa remains effective as long as decarboxylation to dopamine can take place. Levodopa could theoretically lose its efficacy if postsynaptic dopamine receptors disappeared in advanced disease but there is no evidence to suggest they do.

There are two ways which might explain apparent rather than real levodopa insensitivity in advanced Parkinson's disease:

• First, patients in later stages of the disease become more and more likely to develop motor fluctuations. It thus becomes increasingly difficult to prescribe an adequate dose since patients taking higher doses are ‘punished’ with more severe dyskinesias. This leads to the prescription of relatively low and at least partially subtherapeutic doses.

• Second, in advanced disease, most disability arises from degeneration of non-dopaminergic brain regions19 20 causing, for example, cognitive decline, sleep disorders, pain and autonomic dysfunction. Importantly, these non-dopaminergic symptoms greatly determine the quality of life in advanced Parkinson's disease.21 Most of these non-dopaminergic complications cannot be controlled satisfactorily with levodopa. Indeed, optimising dopaminergic therapy in the end phase has been likened to ‘beating a dead horse’.22 Anticholinergic, antipsychotic and antidepressant drugs can be helpful, but novel non-dopaminergic therapies will be required to manage these important problems adequately.9

Argument 4: early treatment with dopamine agonists (and delay of levodopa) postpones motor fluctuations

Fearing the development of levodopa induced motor fluctuations, many clinicians prefer to start symptomatic treatment with a dopamine agonist. Because of their longer elimination half life, agonists should, in theory, provide a more continuous and ‘natural’ stimulation of striatal dopamine receptors.6 Several trials have studied the effectiveness and adverse effects of initial levodopa therapy versus initial dopamine agonists (pergolide, ropinirole or pramipexole).23,–,25 The conclusions were similar. In all studies, levodopa monotherapy was more effective in controlling symptoms and improving activities of daily living. However, this therapeutic advantage had a downside because levodopa monotherapy was associated with earlier, and more severe, motor fluctuations.23,–,25 It is worth looking at the risk of developing severe disabling dyskinesias within the first 5 years of treatment. In one trial comparing levodopa to agonists, this risk was significantly lower in the ropinirole group, regardless of whether or not the patient received supplementary levodopa.23 Similar findings were reported in a study of levodopa versus pramipexole.24 However, after adjusting for disease duration and levodopa dose, the incidence of dyskinesias after initiating levodopa was not significantly different to those initially randomised to levodopa and those initially randomised to pramipexole.26 Furthermore, in all trials, patients on dopamine agonists had a higher risk of short term adverse effects such as ankle oedema and visual hallucinations.

There is a persistent misunderstanding that the effectiveness of levodopa declines or even disappears with time

The interpretation of these results is not straightforward. Levodopa provided a better improvement of parkinsonian signs in all studies compared with the dopamine agonists. This means that the dopamine agonists must have been prescribed in a less effective dose. This leaves unanswered what would have happened if the drugs in both treatment arms had been administered in equipotent doses. Specifically, one cannot exclude the possibility that the same motor fluctuations would have appeared in the dopamine agonist group as in the levodopa group.

Another point of concern is the long term course of patients on dopamine agonist monotherapy. After all, early treatment with a dopamine agonist should in theory result in a structural benefit leading to an enduring absence of (severe) motor fluctuations in the majority of patients. There is currently no convincing evidence to support such a long lasting benefit:

• First, only a minority of patients were able to remain on dopamine agonist monotherapy for longer than 2–3 years. They required addition of levodopa. Probably only patients with benign disease progression can stay on dopamine agonist monotherapy. The fact that these patients develop fewer motor fluctuations could be simply because they have mild disease, not their treatment.

• Second, in the studies mentioned above, patients who were started initially on dopamine agonist monotherapy should have had a persistent advantage, even if levodopa had to be added at a later stage. This does not seem to have been the case. Most studies showed that patients on agonist monotherapy made a ‘catch up sprint’ from the moment levodopa was added and developed motor fluctuations at an accelerated pace (figure 2). Furthermore, an important study—with an impressive follow-up of 14 years—analysed the long term effect of the dopamine agonist bromocriptine versus levodopa in 782 patients with Parkinson's disease.27 This showed that starting with bromocriptine had no advantages on mortality, motor scores or on the severity of motor fluctuations.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Effect of add-on levodopa on the development of motor fluctuations in patients treated with dopamine agonist monotherapy.

A realistic therapeutic goal is to maximise the ON time and to tolerate dyskinesias

Argument 5: the ultimate treatment goal is an ‘ON’ period without any dyskinesias

The severity of dyskinesias plays a central role in this discussion. As mentioned above, dyskinesias should be differentiated into acceptable dyskinesias (ie, present and noticeable, but not markedly interfering with daily activities or quality of life) versus disabling dyskinesias. It is commonly heard that the goal of therapy is to reach a good ‘ON’ motor phase function, without any dyskinesias. However, once patients have developed motor fluctuations, the dyskinesias are an intrinsic part of the therapeutic effect of levodopa28; they inevitably accompany its therapeutic effect. There are only a few patients in whom dyskinesias appear only at doses of levodopa that are much higher than those that produce an antiparkinsonian response.28 Moreover, although motor fluctuations are present in up to 90% of patients who have been treated with levodopa for 10 years, the risk of developing disabling motor fluctuations is lower than is commonly believed.3 5 28 In a recent study, only 12% of the patients developed disabling dyskinesias after 10 years of levodopa therapy.29

These observations send a message to both clinicians and patients who must realise that in advanced Parkinson's disease the intention to attain an ON period without any dyskinesias at all is nearly impossible with the current therapeutic options. A realistic therapeutic goal is to maximise the ON time and to tolerate dyskinesias.

When to start symptomatic treatment?

When considering this issue, patients and clinicians should be guided primarily by the current functional disabilities experienced by the patients during their everyday activities (eg, employment, hobbies, self-care). Concerns about favourable or unfavourable influences on disease progression should not influence this decision because there is no good evidence one way or the other. If a particular patient experiences substantial difficulties in daily activities, symptomatic treatment is justified and can be started, provided the patient agrees.

Which drug to start off with?

There are various options, all with their own specific pros and cons (figure 3). If patients are over 40 years old, a good option is to start with levodopa monotherapy. If the patient starts to develop motor fluctuations, a dopamine agonist can be added.32 33 An alternative is to start with a dopamine agonist and to gradually increase the dose to the maximum tolerated level. Levodopa can be added as ‘rescue’ treatment if necessary.23 24 Because younger patients seem to develop earlier dyskinesias with levodopa, this second option is likely to be preferable for those below the age of onset of 40 years.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

Most motor fluctuations are treatable, and this algorithm can assist clinicians in their therapeutic decision making. COMT, catechol-O-methyltransferase; MAO-B monoamine oxidase B inhibitors.

For each of these two strategies there is good scientific evidence but there has never been a direct comparison between them. Starting with levodopa is certainly cheaper, and in the short term levodopa is often better tolerated than dopamine agonists. But eventually, regardless of the strategy chosen, most patients need combination therapy of two or more antiparkinsonian drugs in the later stages of the disease, especially when motor fluctuations appear. Crucially, more important than the choice of a specific therapy is to prescribe an adequate dose in such a way that the patient really experiences an improvement in their quality of life.

For mildly affected de novo patients, other drugs to try as monotherapy are monoamine oxidase B inhibitors or amantadine. We will not discuss these two treatments further because the debate in this article concerns patients with considerable disability for whom neither monoamine oxidase B inhibitors nor amantadine are usually helpful.

Regardless of the strategy chosen, most patients need combination therapy of two or more antiparkinsonian drugs in the later stages of the disease