Parkinsonism: differential age-trend in Helicobacter pylori antibody


Dobbs Dr 2 Priory Gardens, Berkhamsted, Hertfordshire HP4 2DR, UK. E-mail:



Parkinsonism is associated with prodromal peptic ulceration. Dopamine antagonists provoke experimental ulcer, dopaminergic agents protect, and might inhibit growth of Helicobacter pylori.


To describe the relationship between H. pylori serology and parkinsonism.


Serum H. pylori anti-urease-IgG antibody was measured in 105 people with (idiopathic) parkinsonism, 210 without, from same locality. None had received specific eradication therapy.


Controls showed a birth-cohort effect: antibody titre rose from 30 to 90 years (P < 0.001). Parkinsonism obliterated this (disease status · age interaction, P < 0.05), the differential age trend not being attributable to social class. Those with diagnosed parkinsonism were more likely to be seropositive (odds ratio 2.04 (95% CI: 1.04, 4.22) P < 0.04) before 72.5 years. Overall, titre fell (P=0.01) by 5 (1, 9)% per unit increase in a global, 30-point rating (median 14 (interquartile range 10.5, 17)) of disease severity. No individual category of anti-parkinsonian medication (92% taking) had a differential lowering effect.


Higher prevalence of seropositivity in parkinsonism, before 8th decade, may be due to host susceptibility/reaction, or, conversely, infection with particular H. pylori strain(s) lowering dopaminergic status. Absence of a birth cohort effect in parkinsonism, despite similar social class representation, may be consequent on eradication, spontaneous (gastric atrophy) or by anti-parkinsonian medication.


In 1961, Shwab reported that ‘a number of active ulcers and ulcer-like syndromes were encountered’, in his patients with Parkinson’s disease, ‘suggesting their gastro-intestinal tracts are more vulnerable than in ordinary people of the same age’.1 Strang, in 1965, found that 15% of 200 consecutive out-patients with Parkinson’s disease had been diagnosed as having a peptic ulcer (surgically or radiographically confirmed), compared with 4% of controls (contemporaneous, consecutive out-patients without parkinsonism).2 Approximately three-quarters of the ulcers, in each group, were duodenal. The prodromal ulceration showed ‘very little tendency to chronicity’, and the (antacid/anticholinergic/surgical) treatment ‘almost invariably proved to be definitive’. Strang ascribed the rarity with which the diseases occurred concomitantly (1%) to (anticholinergic) treatment for parkinsonism. This might apply more so now, since dopaminergic agonists and promoters, such as monoamineoxidase-B inhibitors, act, both locally and in the central nervous system, to protect against experimental ulcers.3[4]–5 Indeed, dopaminergic agents have been used clinically in the treatment of duodenal ulcer.6

There appears to be an inverse relationship between central dopaminergic activity and susceptibility to peptic ulcer in experimental models: dopamine antagonists and inhibitors are ulcer-promoting.7[8]–9 The association of previous peptic ulceration with an earlier onset of parkinsonism2 may represent more rapid dopamine depletion than in parkinsonism without this prodrome. Moreover, ulcers are uncommon in schizophrenics, who have dopaminergic overactivity.3 Szabo, in 1979, raised the possibility of whether parkinsonism and duodenal ulcer are both ‘at least partly due to limited exposure to infectious agents or toxic chemicals which impair dopamine production.’3 Altschuler speculated that infection with Helicobacter pylori might be a cause of idiopathic parkinsonism,10 as well as being contributory to peptic ulceration.11 We have substantiated the fit12 with particular respect to concomitant familial aggregation13 and compatible immunological manifestations.14, 15

We now describe a statistical model for the relationship between H. pylori serology and parkinsonism, taking into account personal, disease and treatment characteristics.



A total of 105 patients (55 men and 50 women), diagnosed as having idiopathic parkinsonism, and 210 controls (107 men, 103 women) were studied. Disease severity on a 30-point global rating scale16 had a median value of 14 (interquartile range 10.5, 17: minimum 4, maximum 25). Functional impairment on a 5-point scale17 had a median of III (range II (32%) to IV (33%)). The patients ranged in age from 40 to 89 years (median (interquartile range) 74 (62–78) years). Controls spanned a similar age range (median 70 (58, 77) years), with, additionally, 25 subjects in the preceding decade. In each group, the median social status (UK Registrar General’s Social Classification: I; II; IIIN and IIIM, where N is non-manual and M manual; IV; and V) was IIIN, with an interquartile range of II–IIIN. Each subject obeyed the inclusion and exclusion criteria, and gave written informed consent to participate. The work had local Ethics Committee approval.

Inclusion and exclusion criteria

Severity of parkinsonism, in the patients, had reached the threshold for routine diagnosis. In confirming the diagnosis, the prerequisite was two or more of the four cardinal signs: brady/hypokinesia, rigidity, tremor and postural abnormality, either anatomical or impaired postural reflexes.18 Those in whom there were reservations about the idiopathic nature of the syndrome were excluded.19, 20

All subjects were Caucasian, had English as their first language, and were from the same locality. None had received an specific antimicrobial/anti-secretory therapy against H. pylori. A diagnosis of inflammatory or neoplastic bowel disease, or history of surgery to the gastrointestinal tract, led to exclusion. All were free from cardiovascular/respiratory symptoms during normal activities. Subjects with known arthropathy, or pain in the joints, with another neurological disease, musculoskeletal disorder, or a muscle strength score21 < 4/5 were excluded. Any of the following led to exclusion: intercurrent or chronic physical illness, recent significant change in life situation, clinical depression or other mental illness, clinical dementia or moderate/severe impairment of short and longer term memory (scored22≤ 8/16), receipt of medication with extrapyramidal side-effects.

Clinical assessment

Age, gender, occupational history and tobacco smoking habit were recorded in all subjects. Additionally, in the patients, time since diagnosis, any anti-parkinsonian medication (specifying time taking the dopamine precursor, levodopa) were recorded, and rating scales of severity16 and functional impairment17 applied.


Specific H. pylori anti-urease-IgG-antibody, directed against high molecular weight, cell-bound enzyme, was measured by enzyme-linked immunosorbent assay (ELISA) (SIA Helicobacter pylori (HM-CAP), Sigma-Aldrich Ltd, Poole, UK). A calibration curve converts light absorbance to an ‘ELISA value’ (EV). Serum samples were randomised between and within assay, and duplicates positioned in mirror image on the ELISA plate. The between-assay coefficient of variation was 6.0, 8.0 and 13.0% at an EV of 5.9, 2.4 and 0.8, respectively. An EV > 2.2 is the recommended cut-point of seropositivity, from indeterminate/seronegative (2.2 – 1.8/< 1.8). The proportion of true positive external quality controls correctly identified (sensitivity) was 0.95 (95% CI: 0.75, 0.99), proportion of true negatives (specificity) 1.00 (0.83, 1.00).

Statistical analysis

A generalized linear model was fitted23 to assess the associations between H. pylori antibody titre and candidate covariates: taking EV as dependent variable in the primary analysis avoided a loss of power due to categorization. A gamma probability distribution was assumed for EV, a log link being used to relate the covariates: neurological disease status (with/without parkinsonism), and personal, disease and treatment characteristics. As a preliminary, a differential age trend in parkinsonism (an interaction between age and disease status) was sought. Fractional polynomial regression was used to capture any non-linear effect of age. Logistic regression analysis was used to estimate associations between antibody status (seropositive or not) and the covariates identified above.

Log transformation of time since diagnosis and time on levodopa was necessary to ensure the validity of assumptions of normality of distribution and equality of residual variance.


Age and cohort effects

Figure 1A illustrates that an increment in H. pylori antibody titre with age was typical of the controls (mean 1.55 (95% CI: 0.72, 2.39)%/year, P < 0.001), but not of parkinsonism (−0.29 (−2.86, 2.35)%/year). Indeed, in parkinsonism, no significant age effect could be captured using fractional polynomial regression.

Figure 1.

. Effect of parkinsonism, with respect to age, on (A) H. pylori antibody titre and (B) antibody status. In (A), the regression of EV (95% CI) on age is shown for those with ( inline image) and without ( inline image) parkinsonism. Values are adjusted as if all subjects were male. In (B), the odds ratio for seropositivity (subjects with, P, to those without, C, parkinsonism) is given according to whether age < 72.5 or ≥ 72.5 (see intersection of regression lines in (A)). Odds, attributed to parkinsonism, are after adjustment for subjects’ gender and age in years.

The best generalized linear model for EV contained this differential age trend between patients with parkinsonism and controls (interaction of age with disease status, P < 0.05), age and disease status themselves (P < 0.001 and < 0.05, respectively) and gender (P=0.01). Figure 1A shows that the mean titre of patients exceeded that in controls from 40 (their lower age limit) until 72.5 years (intersection of regression lines). The titre in females was 76.3 (61.5, 94.7)% of that in males, the effect of gender being irrespective of disease status or age. The hypothesis, that inclusion of social class in the best generalized linear model would have no significant effect on EV, was tested (likelihood ratio test, χ2=2.43, d.f.=5, P > 0.7). The differential age trend was not attributable to differences in social class.

Categorization by antibody status (Table 1) showed a greater than expected frequency of seropositivity in parkinsonism, before 72.5 years. After adjustment for gender and years of age, the odds ratio for seropositivity was doubled in these patients ( Figure 1B, P < 0.04). In subjects older than this (131/315), the ratio ranked below, but included, unity. Since only 48% of patients were seropositive, current status is hardly a hallmark of idiopathic parkinsonism. However, might its absence be explained by the natural history of the neurological disease, or by its treatment?

Table 1.  . Frequency table of antibody status by disease status, in 315 volunteers in two age categories Thumbnail image of

Effect of global ratings and duration of parkinsonism

Figure 2 shows the large effect of disease severity on titre. Overall, titre fell by 5.4 (1.2, 9.3)% per unit rise in the 30-point rating (P=0.01), after adjustment for gender. Splitting the rating into four categories (< 10, 10 to < 15, 15 to < 20 and ≥ 20) within the generalized linear model, in order to embrace any nonlinearity, did not improve the fit (likelihood ratio test, χ2=1.84, d.f.=2, P=0.4). There was no significant relationship between time from diagnosis (median (interquartile range) 70 (32, 120) months) and severity rating.

Figure 2.

. Effect of severity (30-point score) of parkinsonism on H. pylori antibody titre. The relationship (95% CI) is shown, after adjustment for gender.

The two global ratings did show some congruity: 47% (adjusted R2) of the variance in severity can be explained by functional impairment. However, the EV was higher (63% (6, 151) after gender adjustment, P=0.03) in mild to moderate disability (stage III) than in minimal (II), but similar (−8 (−42, 46)%, P=0.7) in severe disability (IV) to that in minimal. Time from diagnosis was not different in stages III or IV, to that in stage II.

Neither time from diagnosis, nor the time for which the condition had been judged sufficiently severe to require levodopa (48 (10, 96) months), contributed to the prediction of EV.

Effect of exogenous substances

Categorization of patients simply into those taking anti-parkinsonian medication or not did not help explain the variance in their EV, but only a few were untreated (8%). Titre appeared to be higher (63 (−3, 175)% after gender adjustment, P=0.06) in those treated (83%) with levodopa. However, the lack of levodopa is, of course, a surrogate for minimal functional impairment: the tendency to significance of levodopa in the model was lost on incorporating the functional impairment (but not the categorized severity) rating. Similarly, there was no evidence that other individual drug categories (direct agonist 21%, selegiline 52%, antimuscarinics 16%) influenced EV.

The influence of tobacco smoking was also explored, since it is linked to H. pylori infection,24 but is neuroprotective.25 Non-smokers comprised 46% of all subjects, current smokers 11% and ex-smokers 43%. Antibody titre did tend (P=0.06) to be higher (24.6% (95% CI: −0.01, 56.9)) in smokers, irrespective of whether present or past. This contribution of tobacco smoking habit to the best generalized linear model for EV was independent of disease status (P=0.7 for disease status. smoking category interaction).


A differential age trend in H. pylori antibodies between sufferers from parkinsonism and controls is described. The higher prevalence of seropositivity in parkinsonism (before the 8th decade) may be due to host susceptibility/reaction to infection or, conversely, to infection with particular H. pylori strain(s) which lower dopaminergic status. (The lifetime risk26 of parkinsonism (1 in 40) does not equate to the overall prevalence of H. pylori infection, but then neither does that of peptic ulcer.) The question of epiphenomenon or cause cannot be answered by association. Strength of association is of little help, since the organism is classified only as species when the strain may be relevant.

It might be argued that the differential trend is an artefact. Most infections are acquired in childhood: reference to current social status may mask relatively low parental social status in probands compared with controls, and the susceptibility it conveys. In other words, parkinsonism or its prodrome may give a ‘leg up’ the socio-economic ladder. However, there is no evidence for this. Socio-economic status has not been identified as a risk factor for idiopathic parkinsonism.27 A contemporaneous, detailed case-control study (172 cases and 343 age- and sex-matched controls), in an adjoining county, showed no association between aspects of childhood domestic environment (including size of sibship, position in the birth order, type of housing) and an altered risk.28 As regards upward social mobility, the physical disability itself makes skilled tasks more difficult, the cognitive disability impairs mental agility and ‘multi-tasking’.29, 30 The premorbid personality is obsessional (risk is weakly associated with higher educational level), but depressive and introverted.27

Absence of a birth cohort effect in parkinsonism, despite similar social class representation, may be consequent on the spontaneous eradication of H. pylori (atrophic gastritis, with lightening of the microbial load as gastric pH rises).31 The findings are compatible with a greater destruction of the environment, in which the organism thrives, in more severe parkinsonism and as functional impairment progresses from mild/moderate to severe. If dopaminergic therapy inhibits the growth of H. pylori, increasing dosage may contribute to the fall in antibody titre. Parkinsonism, presenting in older-age, is often relatively quiescent: less virulent strains may be associated with, at most, minimal function impairment.

Upper gastrointestinal dysfunction is widespread in idiopathic parkinsonism,32 and may represent damage to the enteric nervous system. Indeed, there is a loss of myenteric dopaminergic neurones, and Lewy bodies are found in these neurones, as well as in myenteric plexus, and associated sympathetic, ganglia.33, 34

The association between tobacco smoking and H. pylori seropositivity overrides any influences of age, sex and socio-economic status,24 and is shown here to be independent of dopaminergic status, as defined by the presence or absence of diagnosed parkinsonism. The 25% higher antibody titre in smokers, after adjustment for covariates, may, however, constitute their risk factor for peptic ulcer. Being part of a smoking culture might involve an altered exposure to infection or, perhaps more likely, the effect of nicotine on gastric mucosa may facilitate colonization. Any neuroprotective effect of tobacco, for parkinsonism, did not implicate the anti-H. pylori antibody measured. Tobacco smoking lowers circulating cortisol, which is raised in parkinsonism.14 If H. pylori were causally implicated, this neuroprotective effect may be via the inflammatory/(auto)immune response12 rather than the infection itself.

Familial aggregations of parkinsonism are common: approximately 15% of sufferers report one or more relatives with a similar condition.35, 36 Genetics has contributed little, overall, to understanding them.13 Siblings of sufferers from idiopathic parkinsonism, found on screening to share facets of the syndrome (brady/hypokinesia, rigidity, postural abnormality and seborrhoea/seborrhoeic dermatitis), also shared H. pylori seropositivity.13 The odds of being seropositive were three times greater in both sufferers and their siblings than in controls. Mostly, siblings had not lived with proband since the first two decades of life. H. pylori infection is associated with a suppression of the immediate antibody response37 and autoimmunity.38, 39 Its chronicity24, 40 is thus compatible, not only with the long latency of parkinsonism,41 but also with the growing evidence of an active (peripheral and central) pathogenic mechanism which persists beyond its diagnostic threshold.14, 15, 42[43][44][45][46][47][48][49][50][51][52]–53 Moreover, childhood infection40, 54 and later acquisition55 might help reconcile a spectrum, ranging from rapidly progressive parkinsonism in young adults, through facets of the condition quantified in spouses of aged sufferers,56[57][58]–59 to those similarities between what is classified as parkinsonism, and what accepted as ageing.


Dopaminergic status is known to be a determinant of susceptibility to peptic ulcer. The present work suggests that H. pylori infection is implicated in that link, either as a result of host susceptibility/reaction being a function of low dopaminergic status, or certain strain(s) being responsible for dopamine deficiency.

Recognition of gastritis due to H. pylori has revolutionized the therapeutic approach to peptic ulcer disease. The question raised is whether this can be extrapolated to parkinsonism.

H. pylori infection may prove to be an epiphenomenon in parkinsonism. Even so, its closeness of fit for a role as ubiquitous environmental insult may help unravel the aetiology of the neurological disorder.


Clinical work was also carried out at the neighbouring Hillingdon Hospital. We would like to thank Dr R. J. Owen, Head of the Helicobacter Reference Laboratory, Central Public Health Laboratory, London, for his advice, and Drs P. Brady and M. Wilks for carrying out the assay. The work was supported by The Astra Foundation, Mr Harry Bellak of the Harrow Lions Club and the Oak Farm Residents of Hillingdon.