Patients and methods

Consecutive patients receiving oral levodopa with or without oral selegiline (10 mg/day) and satisfying the UKPDS Brain Bank clinical criteria for Parkinson’s disease16 were entered into the study after approval by the hospital ethics committee. The presence or absence of symptomatic autonomic dysfunction did not influence entry. Patients receiving only dopamine agonists or with Hoehn and Yahr stage IV and V disease or age over 75 years were excluded because of concern that their frailty would prevent adequate performance of the tests. Those living beyond metropolitan London were excluded because of the impossibility of attending before noon. All patients taking levodopa or levodopa and selegiline in our clinic seen in a four month period were approached and all entered and completed the study. At the first examination, patients completed a detailed questionnaire of the following symptoms of autonomic dysfunction: postural and postprandial dizziness/impairment of conscious state, impotence, urinary urgency, frequency and incontinence, diminished tolerance to extremes of temperature, Raynaud’s phenomenon/cold extremities, abnormal sweating, constipation, dry eyes, or dry mouth. Those not receiving selegiline were tested once. Those on selegiline were tested once on the drug and three months after its withdrawal. The severity of parkinsonism was rated at each visit (Hoehn and Yahr, North Western University Disability Scale [NWUDS], Websters).

Subjects were studied in a dedicated autonomic function laboratory before noon after taking their normal breakfast and medications. After 20 minutes supine during which patients were familiarised with the equipment and their blood pressure and heart rate had stabilised, autonomic testing using previously described techniques17measured blood pressure and heart rate when supine at 20 minutes and with the following manoeuvres dependent on sympathetic function: head up tilt to 45° at two and 10 minutes and on standing at two minutes, the 30:15 ratio, the response to mental arithmetic for two minutes, application of a cold face pack for 45 seconds, and the cold pressor test for 40 seconds. Vagal function was assessed by the heart rate response to deep breathing for two minutes (R-R variation) and the Valsalva ratio (15 seconds; minimum pressure 20 mm Hg) examined the baroreflex, which is dependent on sympathetic and vagal function. Patients were continuously monitored with a three lead ECG. Blood pressure and heart rate were measured intermittently with a Critikon Dinamap 1846SX. The QT interval was measured manually from a 30 second ECG strip taken after 20 minutes supine so as to detect coincidental predisposition to any arrhythmias which might arise during the study. The humoral response to head up tilt, also dependent on the sympathetic system, was examined. A 16G venflon catheter was inserted before testing and 5 ml blood was taken after 20 minutes supine and 10 minutes tilting for plasma catecholamine concentrations . Samples were immediately mixed with 0.1 ml EGTA/reduced glutathione and stored on ice before centrifugation (3000 rpm for 10 minutes at 4°C). Plasma was pipetted off, immediately frozen, and stored before measurement of noradrenaline and adrenaline concentrations with high pressure liquid chromatography (HPLC) with electrochemical detection.18Comparison between patient groups was with Student’s ttest and the significance of changes in each index induced by each challenge was assessed with one way analysis of variance (ANOVA) (Excel 4.0).

Results

Nine patients on levodopa (group I; three men, six women) and 16 patients taking 10 mg/day selegiline and levodopa (group II; nine men, seven women) were matched for age (group I: 69.6 (SEM 2.3) years; group II: 68 (SEM 1.1), disease duration (group I: 7.1 (SEM 0.8) years; group II: 9.6 (SEM 0.8)), disease severity, and daily levodopa dose (table4). The mean duration of selegiline treatment was 6.4 (SEM 0.8) years. Single patients in each group were taking an ergolene, amantadine, or an anticholinergic or antidepressant drug. These drugs were not changed during the duration of the study. Five group I (56%) and seven group II (44%) patients were participating in the UKPDRG trial. There were no differences between these and the non-trial patients with respect to disease severity or duration, age, frequency of postural dizziness, or antiparkinsonian medications. Three patients (two in group I, one in group II) had treated hypertension. No patient had symptomatic coronary artery disease or risk factors for myocardial ischaemia. All complained of constipation and a dry mouth. Five group I (56%) and six group II (38%) patients reported postural dizziness in the questionnaire, but this was pronounced only in four group II patients. One group II patient had frequent orthostatic hypotensive transient ischaemic attacks suggesting impairment of posterior cerebral circulation (bilateral blindness, vertigo, dysarthria, and impaired consciousness) which developed shortly after he started selegiline; multiple bilateral posterior cerebral watershed infarcts were detected on MRI, but no vertebrobasilar or carotid stenoses were identified by MR angiography. Four group I (44%) and eight group II (50%) patients had urinary urgency and frequency consistent with detrusor instability. None of the patients had clinical or laboratory features of multiple system atrophy or autonomic failure.19

No group I patient developed symptomatic hypotension with head up tilt (table 1, figure). By contrast, selegiline therapy was associated with severe and often symptomatic systolic hypotension on tilting (table 1, figure). In nine (56%) selegiline patients, systolic blood pressure fell by 20% or more after tilting for 10 minutes, whereas only two (22%) group I patients recorded a 20% drop (figure). Head up tilt caused loss of consciousness and hypotensive seizures with an unrecordable blood pressure in two patients on selegiline (figure), only one of whom had a history of postural dizziness. The other three patients with disabling symptomatic postural dizziness before the study were very hypotensive on tilting, although with only mild symptoms. Tilting caused considerable systolic hypotension with severe dizziness and impaired consciousness or cognition in a further four patients taking selegiline, none of whom had a history of postural dizziness. Therefore, five of the six patients with severe hypotension on tilting had no prior orthostatic symptoms. No patient with symptomatic systolic hypotension had bradycardia suggestive of a vasovagal attack.20 All patients who became dizzy or lost consciousness were symptomatic within three minutes of tilting. No selegiline patient with symptomatic hypotension during tilting was symptomatic on standing, even in the presence of frank hypotension. Severe hypotension on tilting was not related to low supine blood pressure. Diastolic blood pressure was variably affected by tilting and standing (tables 1, 2) and was substantially reduced only in the presence of symptomatic systolic hypotension. Hypotension on tilting was associated with a variable and insignificant increase in heart rate and the normal rise in plasma noradrenaline, which was detected in group I, was absent (table 1). After withdrawal of selegiline, head up tilt did not result in hypotension in any patient, including those who were previously hypotensive and symptomatic, and the normal rise in plasma noradrenaline was restored (table 1, figure).

• Download figure
• Open in new tab
• Download powerpoint

Effects of lying, head up tilt at 45⁰, and standing on systolic blood pressure (supine: su, 2 minute tilt: t2, 10 minute tilt: t10, standing: st). Selegiline therapy was associated with orthostatic hypotension on tilting at 10 minutes and lesser hypotension on tilting for 2 minutes and standing. On tilting for 10 minutes, six patients on selegiline developed symptomatic hypotension and in five the blood pressure fell to below 100 mm Hg. Withdrawal of selegiline abolished symptomatic postural hypotension on tilting. The systolic blood pressure at tilting for 10 minutes fell below 100 mm Hg in only one previously severely hypotensive patient after selegiline was stopped and this fall was asymptomatic.

Standing at two minutes had no effect on blood pressure or heart rate in group I patients, but caused an asymptomatic fall in systolic and diastolic blood pressures accompanied by a rise in heart rate in group II patients which was abolished by stopping selegiline (table 2). No other abnormalities of autonomic function were detected (table 3). One patient taking selegiline had 8 beats of ventricular tachycardia (VT) during deep breathing, but a subsequent 24 hour ECG and echocardiograph were normal. Selegiline therapy was not associated with an increased QT interval, even in the patient with ventricular tachycardia (group I: 0.37 (SEM 0.01) s; group II: 0.36 (SEM 0.01) s on selegeline, and 0.38 (SEM 0.01) s after withdrawal).

Stopping selegiline abolished the posterior circulation symptoms in the one patient with complicated orthostatic hypotension, and considerably diminished or abolished the postural symptoms in all previously affected group II patients (table 4). The mean supine systolic and diastolic blood pressures fell after selegiline withdrawal, but this was not significant. Withdrawal of selegiline caused a severe and disabling decline in motor function in 13 (82%) patients. The daily levodopa dose (mg) was generally increased (table 4), although this was not significant, and seven (44%) patients required pergolide and in one case apomorphine to restore comparable motor function. Within two weeks after selegiline was stopped, seven (44%) patients developed transient dysphoria and one (8%) euphoria. Cognition was improved in two (16%) patients and hallucinations disappeared in the sole patient with this symptom.

Discussion

Sustained selegiline therapy over several years was associated with severe and selective systolic hypotension on head up tilt in nine of 16 patients. The diastolic blood pressure was affected only in the presence of severe symptomatic systolic hypotension, whereas there was little tachycardia. Withdrawal of selegiline abolished orthostatic hypotension on head up tilt. The mean supine systolic and diastolic blood pressures fell after selegiline withdrawal, but this was not significant. In a subsequent study (in preparation) we have found a similar fall in supine systolic and diastolic blood pressure which was significant and which we have tentatively interpreted as consistent with a supine pressor effect of selegiline, but the significance and cause of this finding remains uncertain. No patient had multiple system atrophy and global autonomic failure was excluded by clinical and laboratory investigation, including examination of other cardiovascular reflexes. Thus cryptic autonomic failure or drug interactions did not cause orthostatic hypotension. Volume depletion is unlikely to have contributed as all patients were examined in the morning after a normal breakfast and maintained fluid intake before testing. Severe symptomatic selegiline induced hypotension has been reported in multiple system atrophy21 and is a recognised complication of selegiline therapy in Parkinson’s disease.7 8Selegiline was recently reported to be associated with hypotension on head up tilt in Parkinson’s disease.22

The study was open and all eligible patients on levodopa alone or levodopa and selegiline seen in our clinic over four successive months were recruited, regardless of their symptomatology; all completed the study. Thus possible selection bias in the study design was minimised. A potential criticism is that nine patients were recruited in group I versus 16 in group II. However, the two groups were well matched for age, disease severity, and disease duration, although the daily dose of levodopa and incidence of postural dizziness in group I was greater. Postural hypotension resolved after withdrawal of selegiline even though the daily dose of levodopa was generally increased and potentially hypotensive agonists were prescribed in seven patients (44%). Thus it is unlikely that the difference in numbers reflects a selection bias towards more pronounced underlying autonomic dysfunction in patients placed on selegiline. At the time of the study, which was immediately after publication of the UKPDRG trial,4 most patients had been placed routinely on selegiline and levodopa for some years as a result of the initial DATATOP findings.3 We think that this accounts for the difference in the size of the groups.

The mechanism of the hypotensive effect of selegiline is unclear. Few patients took drugs other than levodopa, excluding a drug interaction other than with levodopa. We have subsequently found a similar hypotensive effect in a patient on selegiline monotherapy. None the less, it is not possible to determine if the hypotension found by us was due to selegiline alone or to an interaction with levodopa. Maintenance of systolic blood pressure during passive tilt is thought to be dependent on cardiac output and total peripheral vascular resistance.23 Cardiac output is dependent on heart rate, venous return, and cardiac contractility.24 Heart rate rose equally in group I and group II patients before stopping selegiline and was unaffected by head up tilt in group II patients after drug withdrawal. These results imply that cardiac contractility was impaired in those on levodopa and selegiline, assuming that total peripheral resistance and venous return, which were not measured, did not fall precipitously. Plasma noradrenaline was increased in response to head up tilt in group I and in group II after withdrawal of selegiline, but fell in those receiving selegiline. The rise in plasma noradrenaline in those not receiving selegiline was not significant, even though typical of the normal response for our laboratory. However, as suggested by the large SEM, there were considerable differences in concentrations between patients which may have masked a real physiological effect of head up tilt on plasma noradrenaline with such small numbers. Standing for two minutes caused a small but symptomatic fall in systolic and diastolic blood pressure accompanied by a rise in heart rate in group II patients which was abolished by stopping selegiline and which was not seen in group I (table 2). Presumably, the fall in systolic blood pressure was due to the same effects on cardiac output as occurred with tilting. Diastolic blood pressure after standing for two minutes is maintained by increased sympathetic activity23 as evidenced by the rise in heart rate. Overall, these results suggest that selegiline in combination with levodopa was associated with impaired cardiac output, whereas sympathetic function was probably also impaired as failure of heart rate and plasma noradrenaline to increase with hypotension is thought to indicate autonomic failure.17

The techniques used in this study do not discriminate between the peripheral efferent, peripheral afferent or central components of the autonomic nervous system. Dysfunction of each part of the autonomic system, all of which are affected in Parkinson’s disease,25-29 may cause orthostatic hypotension. Thus, the level(s) of the autonomic system involved in systolic hypotension induced by selegiline in combination with levodopa are uncertain. Similarly, the mechanism(s) by which selegiline might induce orthostatic hypotension is unknown. Non-selective MAO inhibitors which inactivate both isoenzymes and which are not metabolised to amphetamines, cause orthostatic hypotension,30 31possibly because of inhibition of tyramine metabolism.31Short term studies of selegiline at doses of 10 mg/day found that the drug only mildly enhanced the pressor effects of tyramine,1 32 consistent with partial blockade of MAOA and pronounced inhibition of MAOB. Long term exposure to selegiline has been associated with progressively increasing inhibition of MAOA in rats33-35 and humans36 raising the possibility that a similar effect due to increased plasma tyramine could have resulted from chronic therapy at 10 mg/day. Selegiline is metabolised to met-amphetamine and amphetamine.11 The psychiatric symptoms in 52% of patients after drug withdrawal and previous evidence in humans33 and rats33 38-40 suggest that selegiline exerts an amphetamine like effect in vivo. The selective hypotensive effects of selegiline on systolic blood pressure found in this study were similar to those of amphetamine and met-amphetamine in human volunteers.10 13 Chronic exposure to amphetamine also results in tachyphylaxis of the pressor effects of the drug,41 reduces cardiac output in dogs,42reduces striatal dopamine and tyrosine hydroxylase in humans,43 and is neurotoxic to nigrostriatal dopaminergic terminals in experimental animals.44-46 Selegiline has been reported to be neurotoxic to cultured dopaminergic neurons.47 It is not known if these drugs are similarly toxic to sympathetic neurons in humans.

In the UKPDRG study, the mortality of those on selegiline in combination with levodopa began to increase after three years,4 whereas detailed studies of the toxicology of selegiline have been for shorter periods.1 7-9 37 The cause of the increased mortality in the UKPDRG study has not been identified,4 but further information will be available after completion of the cause of death examination (CODE) study. One patient in the present study with recurrent orthostatic hypotensive strokes was clearly at risk of selegiline induced mortality and another with ventricular tachicardia may have been so. If patients with orthostatic hypotension on tilting are included, 56% of the patients in this series had morbidity attributable to selegiline. Clinical studies of selegiline have shown a low incidence of symptomatic orthostatic hypotension, but did not assess changes with head up tilt and plasma concentrations of amphetamine and met-amphetamine were estimated only indirectly via urinary concentrations which were comparable with those seen after oral amphetamine ingestion.7-9 30 Our finding that standing blood pressure and prior orthostatic symptoms are poor indicators of hypotension induced by head up tilt suggests that the hypotensive actions of selegiline may have previously been underestimated.7 8 As amphetamines cause cardiac arrhythmias, intracranial haemorrhage and occlusive cerebral vasculitis in normal humans,48 an amphetamine effect may indicate an increased risk of cardiovascular or cerebrovascular toxicity. Experience of amphetamines in parkinsonism is extremely limited, but benzedrine sulphate therapy over weeks variably affected supine blood pressure (“increased, decreased, unchanged, or vacillating”) in postencephalitic parkinsonism49 and a two week course of amphetamine modestly affected supine blood pressure, but was not reported to cause postural dizziness, in Parkinson’s disease.50

The relation between our finding that selegiline in combination with levodopa has cardiovascular toxicity and the increased mortality associated with this drug regimen in the UKPDRG trial4 is uncertain. In DATATOP,5 6 which did not examine mortality as a primary endpoint or report selegiline induced mortality, patients received levodopa only when their symptoms were uncontrolled by the trial drugs. The projected median time at which DATATOP patients randomised to selegiline alone required levodopa was 2.02 years, which is substantially less than the duration of selegiline therapy in our patients (mean 6.4 years). Thus the patients in our study were not typical of those in the DATATOP trial. Whether or not the rare patient with Parkinson’s disease on selegiline monotherapy after three years is at risk of orthostatic hypotension is unknown. Withdrawal of selegiline resulted in a severe symptomatic motor decline in 82% of patients necessitating increased dopamimetics and confirming a pronounced levodopa sparing effect.7 8 51 Thus the UKPDRG4 and DATATOP5 6 studies and our results are consistent with selegiline acting in Parkinson’s disease only by alleviating motor symptoms.

Conclusion

We have shown that the risks of selegiline in combination with levodopa causing orthostatic hypotension in Parkinson’s disease have been underestimated. Whether or not selegiline alone has similar cardiovascular toxicity is uncertain. It seems prudent to withdraw selegiline from those with symptomatic postural hypotension or concommitant cardiovascular or cerebrovascular disease. For those already on selegiline without appreciable symptomatic morbidity, but with a greater than 20 mm Hg fall in blood pressure on standing for two minutes, gradual withdrawal with a concommitant retitration of levodopa requirements should be considered.