The Lancet: Volume 354, Number 9175 24 July 1999 Possible relation of atypical parkinsonism in the French West Indies with consumption of tropical plants: a case-control study Dominique Caparros-Lefebvre, Alexis Elbaz, and the Caribbean Parkinsonism Study Group* *Members listed at end of paper ---------------------------------------------------------------------------- ---- Department of Neurology, Centre Hospitalier Universitaire (CHU) des Antilles et de la Guyane, Route de Chauvel, 97159 Pointe à Pitre, Guadeloupe, French West Indies (D Caparros-Lefebvre MD); and Inserm U360, Hôpital de la Salpêtrière, Boulevard de l'Hopital, 75013 Paris, France (A Elbaz MD) ---------------------------------------------------------------------------- ---- Correspondence to: Dr Dominique Caparros-Lefebvre (e-mail: [log in to unmask]) Summary Background In Europe and North America, Parkinson's disease is the major form of parkinsonism; less than 4% of cases are progressive supranuclear palsy (PSP) and about 20% are atypical parkinsonism. The distribution of these subgroups is different in the French West Indies. We aimed to define the clinical and demographic specificity of these disorders in Guadeloupe and to investigate a postulated link with consumption of herbal tea and fruits from the Annonaceae family (Annona muricata and Annona squamosa), which contain neurotoxic benzyltetrahydroisoquinoline alkaloids. Methods Between September, 1996, and August, 1998, 87 consecutive patients with parkinsonism were referred to the single neurological department in Guadeloupe. After detailed clinical, neurophysiological, cognitive, and neuroradiological assessment, they were classified by generally accepted criteria as having Parkinson's disease, PSP, or atypical parkinsonism. We compared the amount of tropical fruits and herbal tea consumed by the various parkinsonian subgroups and by frequency-matched controls (patients with benign symptoms and no neurodegenerative disease). Findings Of the 87 patients, 22 had Parkinson's disease, 31 had PSP, 30 had atypical parkinsonism, and four had atypical parkinsonism associated with motor neuron disease, 44 of the patients with PSP or atypical parkinsonism were male. The patients with atypical parkinsonism had symmetrical rigidity and bradykinesia, and no levodopa peak-dose dyskinesias. Patients with PSP differed from those with atypical parkinsonism because they had supranuclear vertical down-gaze palsy, severe gait and balance problems, and frontal-lobe syndrome. 29 patients with PSP reported regular consumption of pawpaw fruit, and 26 drank herbal tea. 30 patients with atypical parkinsonism reported regular consumption of pawpaw fruit, and 24 drank herbal tea. Both of these groups consumed significantly more fruit and herbal tea than patients with Parkinson's disease (fruit: odds ratio 23·6; herbal tea: 28·2); and controls (fruit: 20·7; herbal tea: 6·48). Interpretation Our study confirms the over-representation of atypical parkinsonism and PSP in patients with parkinsonism in the French West Indies. Chronic exposure to neurotoxic alkaloids could be an important aetiological factor because these compounds induce parkinsonism in animals. A larger epidemiological study, to clarify the link between these fruits with atypical parkinsonism and PSP, is proposed. Lancet 1999; 354: 281-86 Introduction In Europe and North America, idiopathic Parkinson's disease is the main cause of parkinsonism, and atypical parkinsonian syndromes account for only 20-30%.1 Clinical and pathological studies have led to the development of operational clinical diagnostic criteria to distinguish between these diseases.2 Classic fully established progressive supranuclear palsy (PSP) accounts for less than 4% of all parkinsonism cases,3 although atypical and incomplete presentations, many labelled provisionally as parkinsonism or atypical parkinsonism, are now well known.4 Sporadic cases of parkinsonism with amyotrophic lateral sclerosis, or combined Parkinson's and Alzheimer's disease are recognised, but remain infrequent except in Guam.5 A 1-year clinical study (August, 1995, to August, 1996) in Guadeloupe, French West Indies, showed that there was an unexpectedly high percentage of atypical parkinsonism unresponsive to levodopa. This island has 420 000 inhabitants and only one department of neurology, where the study was done. Some patients presented with fully established classic PSP, whereas others showed features of Parkinson's disease but with many atypical features. The range of cases closely resembled the descriptions from another study done on the Mariana islands.6 In the French West Indies, PSP and atypical parkinsonism predominated in patients who consumed herbal tea and fruits of the Annonaceae (custard apple or pawpaw family), containing alkaloids that have insecticidal activity.7,8 The herbal tea is used in traditional medicine mainly for its sedative and hypnotic effects. Benzyltetrahydroisoquinolines, present in Annonaceae alkaloids, are neurotoxic to the basal ganglia in animals.9,10 The water-soluble substance reticuline, extracted from leaves, seeds, and bark of these plants, is a major component is a 1 (39-hydroxy-4-methoxybenzyl)-6-methoxy-7-hydroxy-1,2,3,4-tetrahydroisoquinol ine.8 In prospective analysis of all referred parkinsonian patients in Guadeloupe, we assessed the link between ingestion of tropical herbal tea or tropical fruits and the occurrence of atypical parkinsonism and PSP. Methods Patients 87 patients with parkinsonism were referred to the Department of Neurology of the French West Indies University Hospital between September, 1996, and August, 1998. Two-thirds of them were referred by primary-care physicians, and the remainder by neurologists. Exclusion criteria were previous stroke, untreated risk factors for stroke (such as a severe hypertension or diabetes), treatment during the previous 2 years with neuroleptics, history of severe head trauma, or encephalitis. Parkinson's disease was diagnosed by the UK Parkinson's Disease Society Brain Bank criteria.2 The diagnosis of PSP was made with the US National Institute of Neurological Disorders and Stroke (NINDS) consensus criteria, which have nine main requirements:4 a progressive non-familial levodopa-resistant parkinsonism occurring after age 40 years, leading to death within 10 years; postural instability; ophthalmoplegia; pseudobulbar palsy; and frontal-lobe signs. Other clinical features not typical of Parkinson's disease11 were: symmetry of parkinsonian features; predominance of axial rigidity; blepharospasm; eyelid apraxia; palmomental reflex; autonomic failure; bladder incontinence; cerebellar signs; pyramidal signs with brisk tendon reflexes; amyotrophy; and fasciculations. The principal investigator (DCL) examined all patients, and neurologists of the Carribean Parkinsonism study group (PP, ET, AL, GD, MC) examined 52 patients (60% of the cases). An acute levodopa test with 200-400 mg levodopa (or an acute apomorphine test) was done after at least 1 month of treatment with levodopa in line with previous reports.12 Long-term response to levodopa was also assessed. Parkinsonian signs were rated according to the unified Parkinson's disease rating scale.13 Procedures Serum caeruloplasmin and copper studies were done in all patients with atypical parkinsonism, and cerebrospinal fluid was examined for cells staining with periodic acid/Schiff, which suggests Whipple's disease in patients with PSP.14 Electromyography was done to assess involvement of the spinal anterior horn in patients with amyotrophy, and all patients had brain computed tomography or magnetic resonance imaging. Patients underwent cognitive assessment, which included: mini-mental-state examination; Mattis scale; fluency (animals, and "P"); digit span; and the assessment of frontal behaviour (apathy or euphoria, imitation, utilisation behaviour, and grasping reflex). Controls (n=65) were frequency matched for age and sex to the 65 cases with atypical parkinsonism or PSP. Controls were selected from patients admitted for benign symptoms without any neurodegenerative disorder (16 had cardiovascular disease, 16 had trauma, 13 had headache or vertigo, nine had asthma, and 11 had other disorders). All controls were examined by a general practitioner (GM) who was not aware of the specific hypothesis being tested in this study. Reasons for exclusion were signs of parkinsonism or dementia (even minor signs), falls, and hypotension. Patients and controls were asked about their birthplace, place of residence, educational attainment (years of schooling), and consumption of herbal tea and fruits from tropical plants. They were classified into two groups, according to the amount of fruit or herbal tea consumed; those who had eaten fruits or drank herbal tea once a month or more frequently for at least 2 years, and those who had rarely (less than once a month) or never did. More accurate information about frequency and duration of use could not be determined in a comparable way for cases and controls. We assessed consumption of fruits and herbal tea of the Annonaceae; Annona squamosa (common names: pomme cannelle, sweetsop, sugar apple, and cherimoya), Annona muricata (corossol, soursop, guanabana, and graviola), Annona reticulata (cachiman, custard-apple, and mamon). Statistical analysis Basic characteristics of cases and controls were compared with a non-parametric test (Kruskal-Wallis or Wilcoxon test) for comparison of means, and 2 analysis for comparison of proportions. Comparison of fruit or herbal-tea consumption between groups, and calculation of odds ratios and 95% CI was done by logistic-regression analysis, with adjustment for age and sex. Further adjustments for educational attainment did not modify our findings and are therefore not reported. In some cases, logistic-regression models failed to estimate the odds ratio and its CI due to small sample sizes. In such cases, we used the Mantel-Haenszel analysis with adjustment for age (four strata defined by age quartiles) and sex, with a continuity correction of 0·5 in empty cells. This solution is conservative because it underestimates the odds ratio. First, we compared consumption of fruit and herbal tea between cases with aytypical parkinsonism or PSP and cases with Parkinson's disease. Then we compared consumption between cases with atypical parkinsonism or PSP and controls. Comparisons were made for the whole group (atypical parkinsonism and PSP) versus controls, and separate analyses were subsequently done for cases with atypical parkinsonism or PSP. Statistical testing was at the two-tailed level of 0·05. Data were analysed with the SAS package (version 6.12). Results Participants There were 22 (25%) patients with Parkinson's disease, 31 (36%) PSP patients, 30 (35%) patients with atypical parkinsonism, and four (4%) patients with atypical parkinsonism and motor neuron disease (table 1). On average, patients with Parkison's disease were younger than those with atypical parkisonism and PSP; their disease had started at a younger age, and they had a higher mini-mental-state score. Disease duration was not significantly different for both groups. The sex distribution differed slightly but not significantly, among the subgroups. Most of the participants were of Afro-Caribbean or mixed ethnic origin; the remainder were white (two Parkinson's disease, one atypical parkinsonism) or of Indian origin (four Parkinson's disease, six atypical parkinsonism or PSP). A higher proportion of cases with atypical parkinsonism or PSP than of Parkinson's disease cases were of Afro-Caribbean or of mixed ethnic origin. Cases with PSP and atypical parkinsonism were similar to controls in terms of age, sex, educational attainment, and ethnic origin. ---------------------------------------------------------------------------- ---- <SNIP> Table 1: Characteristics of cases and controls --------------------------------------------------------------------------- ----- Characteristics of subgroups Parkinson's disease--Only 22 patients met the criteria for Parkinson's disease,2 with a definite response to levodopa during acute test and over the long term (table 2). The improvement after levodopa therapy varied from 35 to 80%. The median Hoehn and Yahr stage was 2·5 (range 1·5-4). A short-duration levodopa response with fluctuations was noted in all cases and 50% had peak-dose dyskinesias. 17 had a predominant asymmetrical rest tremor, which had been the first symptom. ---------------------------------------------------------------------------- ---- <SNIP> Table 2: The clinical signs and symptoms seen in each subgroup ---------------------------------------------------------------------------- ---- PSP--31 patients had a probable or possible diagnosis of PSP.4 The median Hoehn and Yahr stage was 3·5 (2·5-5). 22 met diagnostic criteria for probable PSP.4 Nine had possible PSP with one major feature missing (absence of gaze palsy in eight cases, presence of familial dementia with parkinsonism in one case). Some of the probable PSP cases had unusual clinical features, particularly a marked mid-brain tremor. Three patients died (two after 6 years, one after 4 years); the cause may have been respiratory failure related to the pseudobulbar syndrome. Bradykinesia and rigidity occurred in all patients, symmetrically in most. All had a predominant axial rigidity, but limb rigidity was variable. Six had additional neck hyperextension. Ten patients had a transient improvement with levodopa therapy, but no sustained response with emergence of motor fluctuations was seen and the acute levodopa test was negative in all. Cognitive testing was possible in 25 patients. Electromyography showed abnormal spontaneous activities in all patients, with polyphasic potentials seen in five patients, and signs of chronic partial denervation, suggesting motor neuron disease. Results of brain imaging were normal in 24 patients and showed moderate subcortical atrophy in seven. Serum copper studies and cerebrospinal fluid were normal in all cases. Atypical parkinsonism--30 patients could not be classified as having Parkinson's disease or PSP (no improvement with levodopa and no supranuclear down-gaze palsy). Rapid progression, a predominantly bradykinetic-rigid syndrome with early balance problems, and little or no response to levodopa without dyskinesias, meant that Parkinson's disease was unlikely.15 The mean duration of levodopa therapy was 2 years (SD 2; range 3 months to 5 years). The median Hoehn and Yahr stage was 3·5 (range 2-5). Four patients had a family history of parkinsonism. Four had dystonia that worsened with levodopa, and involved the limbs in three patients and the neck in one. All but two patients had postural instability that occurred 1-2 years after disease onset, but falls were uncommon during the first year. Eight patients had hypometric saccades. Six patients had an initial positive response to levodopa challenge but the response diminished over time. One had cerebellar signs with ataxia and hypotonia, and cerebellar atrophy indicating a diagnosis of multiple system atrophy. One had severe postural hypotension, which occurred 3 years before parkinsonism, also indicating a diagnosis of multiple system atrophy. During the follow-up, none of the 28 other patients developed typical signs of multiple system atrophy. Cognitive tests were done on 20 patients. Electromyography, done on patients with suspected motor neuron disease, showed moderate to severe denervation and polyphasic potentials. Biological and cerebrospinal-fluid studies were normal. Brain imaging showed bilateral calcifications of internal pallidum in four patients. These calcifications are frequent in people from Guadeloupe and are incidental findings in most cases. Five patients had subcortical atrophy. One patient died after 4 years. Atypical parkinsonism with motor neuron disease--Four patients had associated motor neuron disease suggesting amyotrophic lateral sclerosis, which occurred before onset of parkinsonism. The Hoehn and Yahr stage was 5 in all cases. The parkinsonian features were similar to those in the atypical parkinsonism group. The severity of motor neuron disease masked the akinesia and rigidity at disease onset. Cognitive tests were possible in only two patients. No patients responded to levodopa, and cerebrospinal fluid was normal in all patients. Brain imaging showed frontal atrophy in one patient and was normal in three. Three patients died (one after 4 years and two after 9 months). In one of these patients, neuropathological examination showed predominant changes in the medulla and upper cervical spinal cord, including severe neuronal loss and consequent gliosis in the anterior horns (of the upper cervical spinal cord), nucleus of the XIIth cranial nerve, and nucleus ambiguus. In the remaining neurons of these nuclei there were a few Hirano bodies and many skein-inclusion bodies that were argentophilic and ubiquitin positive. There were no Lewy bodies, and no true neurofibrillary tangles. There was pallor of the pyrimidal tracts more marked in the medullary pyramids and spongiosis of the second layer of the prefrontal cerebral cortex. Occasional skein-inclusion bodies were also found in the nuclei of the reticular formation, locus coeruleus, substantia nigra, and frontal and temporal cortex. In the locus coeruleus and substantia nigra, there was extracellular pigment consistent with mild neuronal loss. These neuropathological findings were similar to those observed in patients with amyotrophic lateral sclerosis or ALS syndrome of Guam.16,17 Follow-up Follow-up ranged from 8 months to 2·5 years. In six patients with PSP and ten patients with atypical parkinsonism, the clinical follow-up suggested an overlap between Parkinson's disease, atypical parkinsonism, and PSP. Indeed, during the first 6 months of treatment, a 25-45% functional improvement with levodopa suggested a diagnosis of Parkinson's disease, even when axial rigidity was prominent and the parkinsonism symptoms were symmetrical in the limbs. After 1-2 years, parkinsonism more closely resembled atypical parkinsonism with a loss of levodopa responsiveness, and later resembled PSP with the emergence of a supranuclear down-gaze palsy. Consumption of tropical fruits and herbal tea Consumption of tropical fruits, herbal tea, or both was reported by higher proportions of patients with PSP or atypical parkinsonism than of those with Parkinson's disease or controls (tables 3 and 4). In separate comparisons with the Parkinson's disease group, the strength of the association with fruits or herbal tea was similar for PSP (odds ratio 5·98 [95% CI 1·05-34·22]) and atypical parkinsonism (6·02 [1·05-34·5]). Adjustment for mini-mental-state examination did not affect these findings. Similarly, in separate comparisons with the control group, the strength of the association with consumption of fruit or herbal tea was similar for PSP (4·35 [1·25-15·2]) and atypical parkinsonism (4·27 [1·22-14·91]). When the two variables fruit consumption and herbal-tea consumption were introduced into the models, the association remained with both of them (fruits 14·0 [3·03-64·8]; herbal tea 4·37 [1·80-10·62]). Conversely, Parkinson's disease was not associated with an increase in consumption of fruit or herbal tea compared with controls (1·20 [0·39-3·75]). Analyses restricted to individuals of Afro-Caribbean origin gave similar results (data not shown). ---------------------------------------------------------------------------- ---- <SNIP> Table 3: Consumption of fruit and herbal tea in cases and controls ---------------------------------------------------------------------------- ---- All patients were asked to stop the consumption of herbal tea and fruits. In two patients with PSP and four patients with atypical parkinsonism, several symptoms improved progressively after they stopped drinking herbal tea, particularly the gait disorder, bradykinesia, and rigidity. All six of these patients were younger than 65 years. In the youngest patient with probable PSP, gait impairment and postural instability were so severe that he was unable to work. Symptoms improved progressively after he stopped daily consumption of herbal tea from Annona muricata, and he was able to return to work after 2 years. The 3-year follow-up showed that 29 patients with atypical parkinsonism or PSP were stable after having been advised to stop consumption of Annonaceae. Of the three patients with atypical parkinsonism and motor neuron disease who died within 1 year, one drank five to ten cups of various herbal teas per day (particularly those from Annonaceae), and two ate two or more fruits of Annona squamosa. ---------------------------------------------------------------------------- ---- <SNIP> Table 4: Odds ratio associated with consumption of fruit and herbal tea ---------------------------------------------------------------------------- ---- Discussion According to generally accepted operational diagnostic criteria,2,4,15 75% of the patients in this study did not have Parkinson's disease. In these atypical cases, many of which resembled PSP, there was a male preponderance. One of the most striking symptoms was the predominance of swallowing and breathing difficulties in patients with atypical parkinsonism, even when amyotrophy and weakness of the limbs were mild. Neuropathological examination was done in only one patient; the findings were similar to those in patients with amyotrophic lateral sclerosis or in the amyotrophic lateral sclerosis subgroup of Guam syndrome, with Hirano bodies and lamellar intracytoplasmic inclusions resembling Bunina bodies.16-18 The frequency of atypical parkinsonism and PSP is higher in Guadeloupe than Europe or USA; these disorders were present in 75% of our parkinsonian patients. We believe that the patients examined provide a reliable picture of the range of parkinsonism in Guadeloupe because most of the island cases are referred at some point to the only department of neurology. We found a strong association between consumption of fruit or herbal tea and PSP or atypical parkinsonism when these patients were compared with cases of Parkinson's disease or with controls. However, our analyses are based on small numbers of cases. The replication of these results on a larger sample of patients is needed. The exposure assessment was mainly qualitative, so we were not able to investigate dose-response relations. The questionnaire evaluated only the 2 years of exposure prior to the diagnosis assessment. A bias cannot be excluded in patients with a longer disease duration who changed their fruit and herbal-tea intake as a consequence of the disease. However, we do not believe that this possibility is likely, because most of the patients had eaten tropical fruits or drunk herbal tea all of their lives and because large differences were observed between Parkinson's disease cases and PSP or atypical parkinsonism cases: there is no obvious reason why PSP or atypical parkinsonism cases would increase their consumption of herbal tea or fruit and Parkinson's disease cases would not, especially since most of PSP or atypical parkinsonism cases had a diagnosis of Parkinson's disease when they were referred to our clinic. The controls were asked about their fruit and tea consumption by a general practitioner who was not aware of the hypothesis under investigation, and cases were interviewed by the neurologist who led the project; however, both investigators attempted to collect information as objectively as possible. These cases bear a close similarity to lytico-bodig of Guam (Andrew Lees personal communication). The overlap between PSP, atypical parkinsonism, and dementia is probably the most characteristic feature, common to Caribbean and Guam syndromes,18 and both have been attributed to a "slow toxin" from certain traditional food.19 Excessive consumption of the sago palm Cycas circinalis has been proposed as a cause of lytico-bodig of Guam.20,21 The possible toxin in these seeds is believed to be ß-N-methylamino-L-alanine.22 Several experimental studies failed to reproduce an animal model of parkinsonism with this compound,23 and the cause of this disorder is unknown. The Annonaceae fruits are commonly used in Guam for their sedative properties.24 A high prevalence of atypical parkinsonism has been also observed in Afro-Caribbean and Indian populations living in England.25 The neurotoxins tetrahydroisoquinolines and 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine have structural similarity and similar physiological properties.26 They are substrates and inhibitors of monoamine oxidase, and inhibitors of mitochondrial complex I.27 The herbal tea from Annonaceae is used daily or weekly by many people in the French West Indies, and it has sedative or hypnotic effects.7,8 These effects could be produced by reticuline (the major component of benzyltetrahydroisoquinolines) which is an opioid precursor,28 and has shown, both in vitro and in vivo, several pharmacological activities including dopaminergic antagonism29,30 and serotoninergic agonism, associated with antidepressive properties.31 The herbal tea of Annona squamosa is consumed mainly by old men, which could explain the male predominance of the disease. This herbal tea is considered to be purgative by some patients, and aphrodisiac by others. The herbal tea of Annona muricata is mainly used as a sedative. None of these plants has been used as antiparkinsonian or antidepressive drugs.7 Quinoline alkaloids have been isolated from tropical plants and also plants that can cause motor neuron disease.32 High concentrations of endogenous benzyltetrahydroisoquinolines have been detected in the cerebrospinal fluid of some parkinsonian patients;33 these substances could be more potent neurotoxins than the methyl derivatives of isoquinoline.34 The cytotoxicity of benzyltetrahydroisoquinolines could be mediated by glutamate35 or free radicals.36 The genetic susceptibility of some individuals to such toxins has been seen previously in the polymorphism of cytochrome P450 activity, which catalyses most oxidative processes.27 In our study, five patients had a family history of neuro-degenerative disorder, which could suggest genetic susceptibility to an exogenous toxin. We suggest that the atypical Parkinsonism seen in many patients in Guadeloupe could be linked to chronic exposure to benzyltetrahydroisoquinolines from tropical herbal teas and fruits. To confirm these findings we are planning a larger case-control study to investigate dose-response relation and the role of specific fruits. Contributors D Caparros-Lefebvre examined all patients, coordinated the linked studies with colleagues, and wrote the paper. A Elbaz did the epidemiological study, the statistical analysis, and wrote the results. The Caribbean Parkinsonism Study Group: Jacqueline Abaul (Faculté Antilles-Guyane), Annick Alpérovitch (INSERM U360, Paris), Alim-Louis Benabid (INSERM U318, Grenoble, France), Valérie Bolivar (CHU Antilles-Guyane), Dominique Charpentier (Department of Ophthalmology, CHU, Pointe à Pitre), Maurice Collard (CHU Strasbourg, France), Gérard Dordain (CHU Clermont-Ferrand, France), Alexis Elbaz (INSERM U360, Paris), Françoise Gray (CHU Paris-Ouest), Reynald Hocquemiller (Laboratoire Pharmacognosie, Châtenay-Malabry, France), Henri Joseph (TRAMIL group), Annie Lannuzel (CHU, Pointe à Pitre), Andrew Lees (National Hospital for Neurology, London, UK), Gwen Mevel (CHU Antilles-Guyane), Charles Pierrot-Deseilligny (CHU Salpêtrière and INSERM U289, Paris), Pierre Pollak (CHU, Grenoble), Sophie Rivaud (INSERM U289, Paris), Catherine Sengler (Pharmaco-Toxicologic Unit, CHU Pointe à Pitre), Eduardo Tolosa (Hopital Clinici Provincial, Barcelona, Spain). <SNIP> References