You are here

  • Home
  • Archive
  • Volume 72, Issue 11
  • Gastrointestinal syndromes preceding a diagnosis of Parkinson’s disease: testing Braak’s hypothesis using a nationwide database for comparison with Alzheimer’s disease and cerebrovascular diseases

Article Text

Article menu
Gut brain axis
Gastrointestinal syndromes preceding a diagnosis of Parkinson’s disease: testing Braak’s hypothesis using a nationwide database for comparison with Alzheimer’s disease and cerebrovascular diseases
Free
  1. Bo Konings1,
  2. Luisa Villatoro2,
  3. Jef Van den Eynde3,
  4. Guillermo Barahona2,
  5. Robert Burns4,
  6. Megan McKnight4,
  7. Ken Hui4,
  8. Gayane Yenokyan5,
  9. Jan Tack1,
  10. Pankaj Jay Pasricha2
  1. 1 Translational Research Centre for Gastrointestinal Disorders (TARGID), KU Leuven University Hospitals, Leuven, Belgium
  2. 2 Department of Medicine, Mayo Clinic Arizona, Scottsdale, Arizona, USA
  3. 3 Department of Cardiology, KU Leuven University Hospitals, Leuven, Belgium
  4. 4 Department of Gastroenterology, Johns Hopkins Medicine, Baltimore, Maryland, USA
  5. 5 Department of Biostatistics, Johns Hopkins University Bloomberg School of Public Health, Baltimore, Maryland, USA
  1. Correspondence to Dr Pankaj Jay Pasricha, Department of Medicine, Mayo Clinic Arizona, Scottsdale AZ 85259, Arizona, USA; Pasricha.Jay{at}mayo.edu

Abstract

Objective Braak’s hypothesis states that Parkinson’s disease (PD) originates in the gastrointestinal (GI) tract, and similar associations have been established for Alzheimer’s disease (AD) and cerebrovascular diseases (CVD). We aimed to determine the incidence of GI syndromes and interventions preceding PD compared with negative controls (NCs), AD and CVD.

Design We performed a combined case-control and cohort study using TriNetX, a US based nationwide medical record network. Firstly, we compared subjects with new onset idiopathic PD with matched NCs and patients with contemporary diagnoses of AD and CVD, to investigate preceding GI syndromes, appendectomy and vagotomy. Secondly, we compared cohorts with these exposures to matched NCs for the development of PD, AD and CVD within 5 years.

Results We identified 24 624 PD patients in the case-control analysis and matched 18 cohorts with each exposure to their NCs. Gastroparesis, dysphagia, irritable bowel syndrome (IBS) without diarrhoea and constipation showed specific associations with PD (vs NCs, AD and CVD) in both the case-control (odds ratios (ORs) vs NCs 4.64, 3.58, 3.53 and 3.32, respectively, all p<0.0001) and cohort analyses (relative risks (RRs) vs NCs 2.43, 2.27, 1.17 and 2.38, respectively, all p<0.05). While functional dyspepsia, IBS with diarrhoea, diarrhoea and faecal incontinence were not PD specific, IBS with constipation and intestinal pseudo-obstruction showed PD specificity in the case-control (OR 4.11) and cohort analysis (RR 1.84), respectively. Appendectomy decreased the risk of PD in the cohort analysis (RR 0.48). Neither inflammatory bowel disease nor vagotomy were associated with PD.

Conclusion Dysphagia, gastroparesis, IBS without diarrhoea and constipation might specifically predict Parkinson’s disease.

  • FUNCTIONAL BOWEL DISORDER
  • APPENDIX
  • BRAIN/GUT INTERACTION
  • IBD
  • ENTERIC NEURONES

Data availability statement

Data are available upon reasonable request. Data exported from TriNetX were saved in Excel files and archived. Every co-author affiliated to Johns Hopkins University was granted access to the TriNetX Research network by the Institute of Clinical and Translational Research (ICTR). The online supplemental materials contain an extensive list of tables representing the original data; researchers will be granted access to the original aggregated data upon reasonable request, with agreement of the corresponding author (PJP).

https://doi.org/10.1136/gutjnl-2023-329685

Statistics from Altmetric.com

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    WHAT IS ALREADY KNOWN ON THIS TOPIC

    • Braak’s hypothesis states that Parkinson’s disease (PD) originates in the gut in a subset of patients, but no studies to date have systematically investigated a broad range of gastrointestinal (GI) symptoms and diagnoses before a diagnosis of PD.

    WHAT THIS STUDY ADDS

    • This is the first multicentre study to establish that dysphagia, gastroparesis, constipation and irritable bowel syndrome without diarrhoea specifically increase the risk of a subsequent new onset diagnosis of idiopathic Parkinson’s disease, even compared with other neurological diseases, such as Alzheimer’s disease and cerebrovascular diseases.

    HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

    • Early detection of GI syndromes might contribute to the identification of patients at risk of PD during a phase were disease modifying therapies could prevent the progression of α-synuclein pathology.

    Introduction

    Parkinsonism is a clinical syndrome characterised by bradykinesia, rest tremor, rigidity and postural instability.1 Its most common cause is Parkinson’s disease (PD), the pathological hallmark of which is thought to be cytoplasmatic eosinophilic Lewy body (LB) depositions. These depositions, mainly consisting of misfolded α-synuclein, have not only been found in the CNS but also in the vagus nerve and enteric nervous system (ENS) of patients with PD.1

    These findings led Braak et al to state the neuroanatomical hypothesis that α-synuclein pathology progresses from peripheral sites such as the ENS to the CNS via vagal or olfactory pathways, thereby introducing the concept that the gastrointestinal (GI) tract might serve as a gateway for environmental factors that induce α-synuclein misfolding and lead to PD.2 A large body of evidence has since then accumulated to support this claim. Even in early untreated stages of the disease, neuropathological studies have found that α-synuclein concentrations in the ENS of patients with PD were higher than those of otherwise healthy individuals, in a characteristic rostrocaudal gradient following visceromotor projections of the vagus nerve.3–5 Complementary studies have shown that various motility disorders5–7 and inflammatory bowel disease (IBD)8 can precede PD and therefore may be risk factors for its development. Moreover, since Gray and colleagues9 first identified the vermiform appendix as a potential source of misfolded α-synuclein, conflicting observational studies have been published about the impact of an appendicectomy on the risk of idiopathic PD.9–14 Finally, two recent registry based studies have strengthened the concept of retrograde vagal α-synuclein propagation by showing that a truncal vagotomy might be protective against the development of PD.15 16 Apart from the bottom up link formulated by Braak, a top down aetiology in which GI symptoms are present in early phases when neurological manifestations are still unnoticed is also supported by experimental evidence.1 Even if no causal link exists, GI syndromes might still represent a risk factor through other mechanisms, or both might be related to a yet unknown third factor.

    Apart from PD, other neurological disorders have also been hypothesised to have GI precedents, either through similar neuroimmune pathways or translocation of microbiome derived neurotoxins into the CNS.17 A strong pathological link between microbiome derived neurotoxins, including Escherichia coli derived Lipopolysaccharide, has been established with disrupted intestinal cell adhesion, impaired synaptic signalling in the Alzheimer’s disease (AD) brain and exacerbation of inflammatory neuropathology.18 19 Additionally, given the prominent role of reactive oxygen species induced inflammation in cerebrovascular diseases (CVD),20 proinflammatory intestinal21 and extraintestinal22 diseases have been linked to a higher risk of CVD than that predicted by conventional risk factors.20

    Previous studies on this topic have been limited by small sample sizes and inadequate controls. Therefore, we used a nationwide electronic health record (EHR) network to investigate the incidence of various GI syndromes and interventions, such as appendectomy and vagotomy, before the onset of PD. Because previous studies lacked specificity for exposures associated with PD, we used a case-control study design to compare patients with PD not only with negative controls (NCs), but also with patients diagnosed with AD and CVD. Additionally, we established a cohort study design for each exposure in the case-control design to validate these findings and establish relative risk (RR) estimates relevant in clinical practice.

    Methods

    Study design and data source

    To investigate the association between various GI syndromes and interventions with the subsequent development of new onset PD, we analysed electronic medical records from the TriNetX Analytics Research Network (Cambridge, Massachusetts, USA). At the moment of data collection, the network consisted of more than 80 million patients from 57 predominantly academic medical centres in the USA. Additional information can be found in the online supplemental methods.

    Supplemental material

    Study population and variables of interest

    In the case control analysis, we examined the incidence of exposures retrospectively (ie, before an initial diagnosis of PD compared with matched controls). Patients with PD were captured using a previously validated method.23 Patients were queried using the International Statistical Classification of Diseases and Related Health Problems, 10th revision (ICD-10) diagnosis of PD (G20), if documented between 1 January 2005 and 1 July 2021; the first ever diagnosis was used as the index event. Only those with at least two prescriptions of an antiparkinsonian drug and a documented ambulatory visit at least 2 years before the first diagnosis of PD were included; secondary causes of PD were excluded. To determine PD specific exposures, control subjects comprised three groups: NCs, and patients with a diagnosis of AD and CVD. NCs consisted of patients without a recorded ICD-10 diagnosis of PD, with at least two documented ambulatory visits between the ages of 50 and 90 years, at least 2 years apart, recorded between 1 January 2005 and 1 July 2021. A minimum of 2 years of retrospective follow-up was ensured by using the second of these visits as the index event. Similarly, 2 years of follow-up was ensured for the AD and CVD groups, and the first ever documented respective ICD-10 diagnosis in the medical records after 1 January 2005 was chosen as the index event. In a pairwise fashion, these groups where then matched to the PD group for age, sex, race and ethnicity using a propensity score matching algorithm.

    To cover the entirety of the GI tract, 18 exposures were investigated: achalasia, dysphagia, gastro-oesophageal reflux disease (GORD), gastroparesis (GP), functional dyspepsia (FD), paralytic ileus (PI), diarrhoea, irritable bowel syndrome (IBS) with constipation (IBS-C), IBS with diarrhoea (IBS-D), IBS without diarrhoea, intestinal pseudo-obstruction (approximate synonym of K59.8: other specified functional GI disorders), faecal incontinence (FI), Crohn’s disease (CD), ulcerative colitis (UC), microscopic colitis (MC), appendectomy and vagotomy. We conducted additional sensitivity analyses in the network, which included stratified analyses based on sex and age at the diagnosis of the index event. A detailed breakdown of the query, inclusion and exclusion criteria, stratified analyses and coding can be found in the online supplemental methods.

    To validate the results from the case-control analyses, we set up a complementary cohort study design. Eighteen cohorts, each diagnosed with one of the investigated exposures in the case-control analysis, were queried and compared with a respective NC cohort (ie, without the exposure) for the prospective risk of developing PD, AD or CVD. Only those with at least 5 years of prospective follow-up were included, and were propensity score matched for age, sex, race and ethnicity, and additionally for a set of potential risk factors and risk modifiers for the development of PD, AD and CVD: arterial hypertension, diabetes mellitus, atrial fibrillation and flutter, and nicotine dependence.

    Statistical analysis

    In the case-control analyses, patients were counted as positive for an exposure if the respective ICD-10 code was documented any time before the first diagnosis of PD or the control health event. To approximate the diagnostic interval between each exposure and the first PD diagnosis, a yearly cross sectional prevalence for each exposure was calculated for the PD and NC groups, up to 6 years before the index event. To detect and quantify potential surveillance bias in our case-control analyses, we collected an agnostic set of negative exposures (Charlson comorbidities). This allowed us to determine the OR that should be considered as indicative of no association. Additionally, we collected positive exposures based on a previous case-control study that identified prodromal motor and non-motor symptoms of PD.24 This enabled us to assess the ability of our dataset to reproduce existing associations. The coding can be found in the online supplemental methods.

    In the cohort analyses, patients diagnosed with the exposure of interest and their NCs were counted as positive for an outcome (PD, AD or CVD) if the respective new onset ICD-10 diagnosis occurred within a 5 year follow-up. Subjects who already had the outcome of interest before the index event were excluded after propensity score matching.

    Exposures and outcomes were collected as absolute numbers; ORs and RRs were calculated with 95% CIs. Standardised mean differences (SMDs) were used to compare baseline characteristics; an SMD of <0.2 was considered well balanced. A Pearson χ2 test was calculated to compare outcomes, and a two sided P value of <0.05 was used to indicate statistical significance. Correction for false discovery rate (FDR) was performed using the step up procedure by Benjamini and Yekutieli, with the Stats package in R (V.4.3.0).25

    Results

    For the case control-study, 24 624 patients with PD met all of the criteria and were matched with 8 267 744 NCs, and 36 187 AD and 528 207 CVD patients, giving 24 624 patients in the comparison with NCs, 19 046 with AD and 23 942 with CVD. Baseline characteristics after pairwise matching are presented in table 1; minimal differences in age at index persisted. SMDs and p values before and after matching can be found in online supplemental tables 1,2.

    View this table:
    Table 1

    Baseline characteristics for subjects with PD and controls in the case-control analyses after pairwise matching.

    The results of the case-control analyses are presented in figure 1 and online supplemental table 3. All GI syndromes were significantly increased in the PD group compared with NCs (OR >1; p<0.05). However, only dysphagia (OR 3.58), GP (OR 4.64), FD (OR 3.39), intestinal pseudo-obstruction (OR 3.01), diarrhoea (OR 2.85), constipation (OR 3.32), IBS-C (OR 4.11), IBS-D (OR 4.31), IBS without diarrhoea (OR 3.53) and FI (OR 3.76) gave ORs that were numerically greater than the upper limit of the negative exposures (OR range 1.20–2.79; online supplemental tables 4–6 and online supplemental figures 1–3). Furthermore, only dysphagia, GP, IBS-C, IBS without diarrhoea and constipation were specific for PD (OR >1; p<0.05) when compared with both the AD and CVD group. After correcting for FDR, GP and constipation did not remain significantly different (p>0.05) compared with the CVD and AD groups, respectively. Other exposures were not only significantly associated with PD, but also showed strong associations with the AD or CVD group. For example, FI appeared to be equally increased before AD, and diarrhoea was even more increased before the onset of both AD and CVD.

    For FD, IBS-D and intestinal pseudo-obstruction, the risk of PD, AD and CVD did not differ significantly (p>0.05). The remaining exposures, including achalasia (OR 1.92), GORD (OR 2.18), PI (OR 2.63), CD (OR 1.99), UC (OR 1.87), MC (OR 2.19) and appendectomy (OR 2.40) showed positive associations with PD compared with NCs, but gave ORs below the upper limit of what is expected by surveillance bias (OR range 1.20–2.79). Only for GORD and appendectomy we observed significant differences compared with AD (OR 1.14, p<0.0001) and CVD (OR 0.57, p=0.03), respectively. The latter did not remain significant after correction for FDR (p=0.21). Prior vagotomy did not impact the risk of PD.

    Figure 1
    Figure 1

    Case-control analyses. Odds ratios (ORs) of previous exposures in patients with Parkinson’s disease (PD) compared with matched negative controls (NCs), or patients with Alzheimer’s disease (AD) and cerebrovascular diseases (CVD). ᵃIn the case-control analysis, the incidence of exposures was examined retrospectively (ie, before an initial diagnosis of PD compared with matched NCs, and AD and CVD patients). Patients were diagnosed with PD or the respective control health event between 1 January 2005 and 1 July 2021, and NCs were queried using two ambulatory visits during the same time window. Patients and exposures were identified with the International Statistical Classification of Diseases and Related Health Problems, 10th revision (ICD-10) codes, and Current Procedural Terminology (CPT) codes, using electronic medical health record data from the TriNetX research network (online supplemental methods). Exposures were included if they were documented any time before the diagnosis of PD or the control health event, in the available medical records. ᵇORs were calculated as follows: odds of documented exposure in the PD cohort/odds of documented exposure in the control cohort. Absolute rates can be found in online supplemental table 3. ᶜP values were calculated using a Pearson χ2 test, after matching for baseline characteristics (ie, age, sex, race and ethnicity). dCorrection for false discovery rate was performed using the step up procedure by Benjamini and Yekutieli, with the Stats package in R. *The term intestinal pseudo-obstruction was used as an approximate synonym for the ICD-10 code K59.8 other specified functional intestinal disorders.**The term IBS not otherwise specified is commonly used as an approximate synonym for the ICD-10 code K58.9 IBS without diarrhoea. IBS, irritable bowel syndrome.

    For GP, women were approximately twice as likely as men to develop PD (OR 7.3 for women, 3.05 for men, both p<0.0001 (online supplemental tables 7–9 and online supplemental figure 4), and the OR for GP was especially high for early onset PD (online supplemental tables 10–11). Exclusion of previous antidopaminergic drug use did not significantly alter any associations with PD compared with NCs (online supplemental table 12 and online supplemental figure 5). The ORs of all PD specific exposures were positioned well within the range of those of established motor and non-motor prodromes of PD (ie, positive exposures; OR PD vs NCs 2.04–7.41). An approximation of the diagnostic interval for each exposure can be found in online supplemental figure 6.

    Figure 2 and online supplemental tables 13–14 show the results of the cohort analyses. A significantly increased RR of new onset PD (RR >1; p<0.05) was found after a diagnosis of dysphagia (RR 2.27), GORD (RR 1.13), GP (RR 2.43), FD (RR 1.15), intestinal pseudo-obstruction (RR 1.84), diarrhoea (RR 1.32), constipation (RR 2.38), IBS without diarrhoea (RR 1.17) and FI (RR 1.74); all except for FD (p=0.09) remained significant after correction for FDR. However, only for dysphagia, GP, intestinal pseudo-obstruction, IBS without diarrhoea and constipation was this RR numerically higher than the RR of developing AD and CVD, and statistical significance (p<0.05) was achieved only for constipation and dysphagia (online supplemental table 15). In line with the case control analyses, FI equally increased the risk of PD and AD (PD: RR 1.74; AD: RR 1.76, both p<0.0001), and in the GORD cohort the RR of developing CVD was higher than that of PD (CVD: RR 1.38; PD: RR 1.13, both p<0.0001). FD and diarrhoea were associated with all three disorders (p≤0.05), and PI, CD and MC only increased the risk of CVD (RR 1.22, p<0.0001; RR 1.20, p<0.0001; RR 1.26, p=0.02, respectively). Although no specific association was found in the case-control analysis, appendectomy significantly reduced the risk of developing PD (RR 0.48, p=0.05), which did not remain significant after correction for FDR. For all other exposures (ie, achalasia, UC, IBS-C, IBS-D and vagotomy), no significant associations were found.

    Figure 2
    Figure 2

    Cohort analyses. Relative risks (RRs) of developing Parkinson’s disease (PD), Alzheimer’s disease (AD) or cerebrovascular diseases (CVD) within 5 years of the diagnosis of a given exposure, compared with negative control (NCs) without the respective exposure. ᵃFor each analysis, a cohort of patients identified by the diagnosis of a given exposure was compared with their respective NCs for the prospective risk of PD, AD and CVD within 5 years of the index event (ie, diagnosis of the given exposure, or a visit for NCs). After propensity score matching, patients that already had the investigated outcome (ie, PD, AD or CVD) documented before the index event were excluded from the analysis. Exposures and outcomes were identified using the International Statistical Classification of Diseases and Related Health Problems, 10th revision (ICD-10,) and Current Procedural Terminology (CPT) codes. Electronic medical health record data were collected from the TriNetX research network. Diagnostic coding can be found in the online supplemental methods. ᵇRRs were calculated as follows: risk of outcome in the exposure cohort/risk of outcome in the control cohort. Absolute rates can be found in online supplemental table 13. ᶜP values were calculated with a Pearson χ2 test, after matching for baseline characteristics and risk factors. Baseline characteristics included age, sex, race and ethnicity; risk factors included arterial hypertension, diabetes mellitus, atrial fibrillation and flutter, and nicotine dependence. dCorrection for false discovery rate was performed with the step up procedure by Benjamini and Yekutieli, using the Stats package in R. *The term intestinal pseudo-obstruction was used as an approximate synonym for the ICD-10 code K59.8 other specified functional intestinal disorders. **The term IBS not otherwise specified is commonly used as an approximate synonym for the ICD-10 code K58.9 IBS without diarrhoea. IBS, Irritable bowel syndrome.

    Discussion

    We used a nationwide EHR network to comprehensively investigate disorders across the entire GI tract before a diagnosis of PD. We used two complementary study designs to establish that dysphagia, gastroparesis, constipation and IBS without diarrhoea specifically increase the risk of a subsequent new onset diagnosis of idiopathic PD, even compared with other neurological diseases, such as AD and CVD.

    Surveillance bias is an inherent problem in observational studies. When not addressed appropriately, it can compromise the validity of causal inference and lead to irreproducible results.26 A broader implementation of empirical approaches to evaluate and correct for the presence of systematic error in observational studies is needed.27 Therefore, we set up an approach to understand the true extent of surveillance bias in our study and its potential contribution to implicating premorbid factors for PD. Hence we collected data on all diagnoses included in the Charlson comorbidity index. These premorbid conditions were considered as a comprehensive set of agnostic negative exposures. In our case-control study, we observed statistically significant increases in most of these exposures in PD cases compared with NCs, but not compared with AD and CVD (online supplemental figure 1). To establish whether these increases represented surveillance bias or true associations (although unlikely based on the current literature), we investigated the same exposures for other neurological disorders (AD and CVD) compared with their NCs (online supplemental figure 3). Since the same significant correlations emerged, surveillance bias was likely. Subsequently, we determined the range of ORs that should be expected if they are a result of surveillance bias alone. These ORs ranged between 1.20 and 2.79 in the analysis of PD with NCs. To determine whether existing associations with PD could be replicated with ORs greater than those of negative exposures, we also collected prodromal motor and non-motor symptoms of PD (ie, positive exposures). These resulted in ORs ranging between 2.04 and 7.41 (online supplemental figure 1).

    Having established a measure of surveillance bias in the case-control study, we then determined ORs for the putative GI pre/comorbidities of PD. Relative to the upper limit of the negative exposures, GI exposures fell into two categories. First were those for which the ORs overlapped with the ORs expected for surveillance bias (OR 1.20–2.79). For these exposures, including achalasia, GORD, PI, IBD (ie, CD, UC, and MC), appendectomy and vagotomy, we cannot be confident that these were true associations, although within the constraints of our study we cannot categorically state that they were not. Second were those for which the ORs were clearly higher than the ORs expected for surveillance bias (OR >2.79). For these exposures, including dysphagia, GP, FD, intestinal pseudo-obstruction, diarrhoea, constipation, IBS-C, IBS-D, IBS without diarrhoea and FI, we can confidently state to have established significant associations with new onset PD. To determine the specificity of the identified significant exposures for PD in the case-control analyses, we subsequently compared subjects with PD with subjects with AD and CVD. Only dysphagia, GP, constipation, IBS without diarrhoea and IBS-C remained specific for PD compared with both neurological diseases (table 2). Importantly, we cannot exclude the possibility that these factors might still be associated with these diseases, although at a smaller scale. Similarly, while other exposures were not specific for PD (ie, FD, intestinal pseudo-obstruction, diarrhoea, IBS-D and FI), we cannot strictly exclude the possibility that these conditions might still be risk factors for PD.

    View this table:
    Table 2

    Summary of Parkinson’s disease specific exposures in the case-control and cohort studies

    Finally, to validate these findings both in terms of their significance and specificity and establish one RR estimate of developing PD after the diagnosis of each exposure, we set up a complementary cohort study. Here, five exposures (ie, dysphagia, GP, IBS without diarrhoea, intestinal pseudo-obstruction and constipation) significantly increased the risk of PD and resulted in RRs that were numerically greater than those of AD and CVD (table 2). Four of these provide internal validation for PD specific exposures identified in the case-control analysis. These exposures are thus very unlikely to be a result of selection or surveillance bias and can therefore be considered the most significant findings from our study. Minor discrepancies can be explained by intrinsic differences in the study designs.

    The consistent correlation between constipation and PD (RR 2.38, 95% CI 2.24 to 2.54) confirms an abundance of existing literature. Previous reports have stated that constipation can even precede PD by up to 20 years.28 More surprising is the strong association for dysphagia (RR 2.27, 95% CI 2.10 to 2.45), which has so far mainly been reported after its diagnosis.1 The prevalence of oesophageal dysmotility in PD has been shown to be as high as 80% when using objective measures,29 but a delay in oropharyngeal transit has also been found in drug naïve and subjectively asymptomatic phases of the disease.30 31 While evidence supports that oropharyngeal function might be affected through brainstem and cortical areas,6 29 post mortem studies by Mu et al showed that pharyngeal muscles, sensory neurons and motor neurons are also often affected by LB pathology in PD.32 33 Interestingly, the highest RR was observed for GP (RR 2.43, 95% 1.92 to 3.09), a disorder characterised by delayed gastric emptying (GE) in the presence of symptoms such as nausea, vomiting, early satiety, postprandial fullness, belching and bloating, and the absence of mechanical obstruction.34 As Braak et al hypothesised,35 the multitude of gastric vagal connections makes GP an especially promising candidate as a biomarker of PD.6

    Furthermore, despite considerable overlap in symptoms and pathology,36 the lack of PD specificity for FD suggests that an established delay in GE in the presence of symptoms is more strongly associated with PD than symptoms alone, assuming that ICD codes are taken at face value. This indicates that objective changes in enteric physiology may provide a more reliable measure for evaluating enteric involvement in PD. Although the prevalence of delayed GE in PD has been reported to range from 70% to 100%,37 reports of GP preceding PD remain anecdotal.37 Because of its relatively low prevalence, it should not be a surprise that our study is the first to provide observational evidence that GP and dysphagia might precede PD.5 6 More established is the association of IBS with the subsequent development of PD.38–40 IBS-C and IBS without diarrhoea were both specifically increased in the case-control analysis, but only the latter was replicated in the cohort analysis (RR 1.17, 95% CI 1.05 to 1.3). Importantly, a Swedish study revealed that although the positive predictive value of ICD-10 codes for IBS is generally high (80–95%), their accuracy in indicating specific subtypes was considerably lower (55–67%).41 Nevertheless, increased intestinal permeability constitutes a core pathophysiological mechanism in a major subset of IBS patients,42 and has also been found in PD.38 Routine colonoscopies for clinically suspect IBS could become important to determine the presence of LB pathology in patients at risk for PD, only if intestinal LB pathology becomes an established biomarker, which until now it has not.39

    Although anorectal symptoms are among the most frequent GI symptoms in PD, our data suggest that the presence of FI might not distinguish between the development of PD and other neurodegenerative diseases.43 Even if not prodromal to AD, our findings support the fact that the progression of cognitive decline in AD is frequently unmasked by FI.44 While diarrhoea and FD increased the risk of all three diseases, intestinal pseudo-obstruction showed specificity for PD in the cohort analyses (RR 1.84, 95% CI 1.18 to 2.87). This disorder, characterised by impaired peristalsis and presumably caused by a neuropathy or myopathy, has been described in various neurological disorders, including PD.45

    Finally, some exposures in the case-control analyses gave ORs in the range of those expected from surveillance bias. These exposures included achalasia, GORD, PI, IBD (ie, CD, UC, and MC), appendectomy and vagotomy. Other than for GORD, the cohort analyses also did not show any significant associations with PD for these exposures. While IBD has been linked to PD in various observational46 47 and genetic studies,48 neither of our study designs supported this link. However, we cannot strictly dismiss the possibility of an association based on this empirical surveillance bias cut-off and a relatively limited follow-up. In addition, we were unable to assess the impact of anti-tumour necrosis factor (anti-TNF) therapy exposure, which has been hypothesised to decrease the risk of PD.49 Prospective studies are necessary to investigate this association and to establish whether anti-TNF therapy can effectively protect against PD. Notably, concordant with an earlier study that linked reflux oesophagitis to an increased risk of stroke and transient ischaemic attack in patients with atrial fibrillation,21 the risk of CVD in our study was significantly greater after a diagnosis of GORD, CD and MC. These findings suggest that a better understanding of the link between GI inflammation and cerebrovascular events may lead to improved risk stratification and identification of new preventive strategies.21

    After Grey et al first discovered that α-synuclein was most abundant in the appendiceal mucosa,9 conflicting evidence has emerged about the impact of an appendectomy on PD risk. While three studies did not find any association,11 14 50 one abstract reported an increased risk of PD13 and two large observational studies supported a protective effect.10 12 Despite our limited follow-up and sample size compared with the aforementioned studies, we observed a relative risk reduction of 52% in our cohort analysis, while the case-control analysis was likely underpowered to detect any consistent association for appendectomy. Multiple studies suggest that the appendix constitutes a prominent source of seeding competent pathologically folded α-synuclein,10 and houses bacteria capable of releasing inflammatory mediators.10 The subsequent migration of α-synuclein to the CNS has been substantiated by studies showing a protective effect of a truncal vagotomy on PD development.15 16 Compared with these reports, our study was underpowered to detect any consistent associations for vagotomy.15 16

    Finally, we attempted to assess the proximity of each diagnosis to the diagnosis of PD in the case-control study (online supplemental figure 5). We found that the OR for dysphagia and constipation decreased considerably as the distance from the diagnosis of PD increased, while the OR for GP and IBS without diarrhoea remained relatively constant. This suggests that differences in lead time exist, but future longitudinal population based studies will be crucial to determine whether these PD specific GI syndromes are part of the early manifestation of PD or truly precede the disease. Importantly, the combination of two complementary study designs reduced the potential for selection bias. The case-control analysis consisted of patients with PD, AD or CVD without the requirement of any previous exposure, while the cohort analyses consisted of patients with newly diagnosed GI exposures without the requirement of a subsequent PD, AD or CVD diagnosis. This study is subject to intrinsic limitations of EHR data, including unknown completeness of records and absent validation of diagnoses. The multicentre character and inclusion of racially and ethnically diverse subjects ensured that these results are generalisable to patients at academic medical centres across the USA.

    Conclusion

    This study is the first to establish substantial observational evidence that the clinical diagnosis of not only constipation, but also dysphagia, GP and IBS without diarrhoea might specifically predict the development of PD, whereas other exposures were less specific. An appendectomy appeared protective, leading to further speculation about its role in PD pathophysiology. These findings warrant alertness for GI syndromes in patients at higher risk for PD and highlight the need for further investigation of GI precedents in AD and CVD. To establish a stronger body of clinicopathological evidence, we advocate for future studies to assess the sensitivity and specificity of these disorders and their clinicopathological correlates for the early detection of neuropathology.

    Data availability statement

    Data are available upon reasonable request. Data exported from TriNetX were saved in Excel files and archived. Every co-author affiliated to Johns Hopkins University was granted access to the TriNetX Research network by the Institute of Clinical and Translational Research (ICTR). The online supplemental materials contain an extensive list of tables representing the original data; researchers will be granted access to the original aggregated data upon reasonable request, with agreement of the corresponding author (PJP).

    Ethics statements

    Patient consent for publication

    Acknowledgments

    BK and JV were supported by a fellowship of the Belgian American Educational Foundation (BAEF). Parts of this work have been presented as a conference abstract at Digestive Disease Week 2022, San Diego, California, USA (doi: 10.1016/S0016-5085(22)60469-4).

    References

    1. ↵
      2. Travagli RA ,
      3. Browning KN ,
      4. Camilleri M
      . Parkinson disease and the gut: new insights into pathogenesis and clinical relevance. Nat Rev Gastroenterol Hepatol 2020;17:673–85. doi:10.1038/s41575-020-0339-z
      OpenUrlCrossRef
    2. ↵
      2. Hawkes CH ,
      3. Del Tredici K ,
      4. Braak H
      . A timeline for Parkinson’s disease. Parkinsonism Relat Disord 2010;16:79–84. doi:10.1016/j.parkreldis.2009.08.007
      OpenUrlCrossRefPubMedWeb of Science
    3. ↵
      2. Lebouvier T ,
      3. Neunlist M ,
      4. Bruley des Varannes S , et al
      . Colonic biopsies to assess the neuropathology of Parkinson’s disease and its relationship with symptoms. PLoS One 2010;5:e12728. doi:10.1371/journal.pone.0012728
    4. ↵
      2. Shannon KM ,
      3. Keshavarzian A ,
      4. Mutlu E , et al
      . Alpha-synuclein in colonic submucosa in early untreated Parkinson’s disease. Mov Disord 2012;27:709–15. doi:10.1002/mds.23838
      OpenUrlCrossRefPubMed
    5. ↵
      2. Cersosimo MG ,
      3. Raina GB ,
      4. Pecci C , et al
      . Gastrointestinal manifestations in Parkinson’s disease: prevalence and occurrence before motor symptoms. J Neurol 2013;260:1332–8. doi:10.1007/s00415-012-6801-2
      OpenUrlCrossRefPubMed
    6. ↵
      2. Fasano A ,
      3. Visanji NP ,
      4. Liu LWC , et al
      . Gastrointestinal dysfunction in Parkinson's disease. Lancet Neurol 2015;14:625–39. doi:10.1016/S1474-4422(15)00007-1
      OpenUrlCrossRefPubMed
    7. ↵
      2. Jost WH
      . Gastrointestinal dysfunction in Parkinson’s disease. J Neurol Sci 2010;289:69–73. doi:10.1016/j.jns.2009.08.020
      OpenUrlCrossRefPubMed
    8. ↵
      2. Brudek T
      . Inflammatory bowel diseases and Parkinson’s disease. J Parkinsons Dis 2019;9:S331–44. doi:10.3233/JPD-191729
      OpenUrl
    9. ↵
      2. Gray MT ,
      3. Munoz DG ,
      4. Gray DA , et al
      . Alpha-synuclein in the appendiceal mucosa of neurologically intact subjects. Mov Disord 2014;29:991–8. doi:10.1002/mds.25779
      OpenUrlPubMed
    10. ↵
      2. Killinger BA ,
      3. Madaj Z ,
      4. Sikora JW , et al
      . The vermiform appendix impacts the risk of developing Parkinson's disease. Sci Transl Med 2018;10:eaar5280. doi:10.1126/scitranslmed.aar5280
    11. ↵
      2. Marras C ,
      3. Lang AE ,
      4. Austin PC , et al
      . Appendectomy in mid and later life and risk of Parkinson’s disease: A population-based study. Mov Disord 2016;31:1243–7. doi:10.1002/mds.26670
      OpenUrl
    12. ↵
      2. Mendes A ,
      3. Gonçalves A ,
      4. Vila-Chã N , et al
      . Appendectomy may delay Parkinson’s disease onset. Mov Disord 2015;30:1404–7. doi:10.1002/mds.26311
      OpenUrl
    13. ↵
      2. Sheriff MZ ,
      3. Mansoor E ,
      4. Cooper GS
      . 739 – Parkinson’s disease is more prevalent in patients with Appendectomies: A national population-based study. Gastroenterology 2019;156:S–1409. doi:10.1016/S0016-5085(19)40573-8
      OpenUrlCrossRef
    14. ↵
      2. Svensson E ,
      3. Horváth-Puhó E ,
      4. Stokholm MG , et al
      . Appendectomy and risk of Parkinson's disease: A nationwide cohort study with more than 10 years of follow-up. Mov Disord 2016;31:1918–22. doi:10.1002/mds.26761
      OpenUrl
    15. ↵
      2. Liu B ,
      3. Fang F ,
      4. Pedersen NL , et al
      . Vagotomy and Parkinson disease: A Swedish register-based matched-cohort study. Neurology 2017;88:1996–2002. doi:10.1212/WNL.0000000000003961
      OpenUrlCrossRefPubMed
    16. ↵
      2. Svensson E ,
      3. Horváth-Puhó E ,
      4. Thomsen RW , et al
      . Vagotomy and subsequent risk of Parkinson’s disease. Ann Neurol 2015;78:522–9. doi:10.1002/ana.24448
      OpenUrlCrossRefPubMed
    17. ↵
      2. Zhong S-R ,
      3. Kuang Q ,
      4. Zhang F , et al
      . Functional roles of the microbiota-gut-brain axis in Alzheimer’s disease: implications of gut microbiota-targeted therapy. Transl Neurosci 2021;12:581–600. doi:10.1515/tnsci-2020-0206
      OpenUrl
    18. ↵
      2. Zhao Y ,
      3. Cong L ,
      4. Lukiw WJ
      . Accumulates in neocortical neurons of Alzheimer’s disease (AD) brain and impairs transcription in human neuronal-glial primary co-cultures. Front Aging Neurosci 2017;9:407. doi:10.3389/fnagi.2017.00407
    19. ↵
      2. Pogue AI ,
      3. Jaber VR ,
      4. Sharfman NM , et al
      . Downregulation of neurofilament light chain expression in human neuronal-glial cell co-cultures by a microbiome-derived lipopolysaccharide-induced miRNA-30B-5P. Front Neurol 2022;13:900048. doi:10.3389/fneur.2022.900048
    20. ↵
      2. Pawluk H ,
      3. Woźniak A ,
      4. Grześk G , et al
      . The role of selected pro-inflammatory cytokines in pathogenesis of ischemic stroke. Clin Interv Aging 2020;15:469–84. doi:10.2147/CIA.S233909
      OpenUrlPubMed
    21. ↵
      2. Jiang Y ,
      3. Damiris K ,
      4. Suero-Abreu G , et al
      . Reflux esophagitis is associated with higher risks of acute stroke and transient ischemic attacks in patients hospitalized with atrial fibrillation: A nationwide inpatient sample analysis. Medicine (Baltimore) 2021;100:e26502. doi:10.1097/MD.0000000000026502
    22. ↵
      2. Liliang P-C ,
      3. Liang C-L ,
      4. Lu K , et al
      . Population-based study suggests an increased risk of Alzheimer’s disease in Sjögren’s syndrome. Clin Rheumatol 2018;37:935–41. doi:10.1007/s10067-017-3940-y
      OpenUrl
    23. ↵
      2. Alonso A ,
      3. Rodríguez LAG ,
      4. Logroscino G , et al
      . Gout and risk of Parkinson disease: a prospective study. Neurology 2007;69:1696–700. doi:10.1212/01.wnl.0000279518.10072.df
      OpenUrlCrossRefPubMed
    24. ↵
      2. Schrag A ,
      3. Horsfall L ,
      4. Walters K , et al
      . Prediagnostic presentations of Parkinson's disease in primary care: a case-control study. Lancet Neurol 2015;14:57–64. doi:10.1016/S1474-4422(14)70287-X
      OpenUrlCrossRefPubMed
    25. ↵
      2. Benjamini Y ,
      3. Yekutieli D
      . The control of the false discovery rate in multiple testing under dependency. Ann Statist 2001;29:24. doi:10.1214/aos/1013699998
      OpenUrl
    26. ↵
      2. Lipsitch M ,
      3. Tchetgen Tchetgen E ,
      4. Cohen T
      . Negative controls: a tool for detecting confounding and bias in observational studies. Epidemiology 2010;21:383–8. doi:10.1097/EDE.0b013e3181d61eeb
      OpenUrlCrossRefPubMedWeb of Science
    27. ↵
      2. Schuemie MJ ,
      3. Ryan PB ,
      4. DuMouchel W , et al
      . Interpreting observational studies: why empirical calibration is needed to correct P-values. Stat Med 2014;33:209–18. doi:10.1002/sim.5925
      OpenUrlCrossRefPubMed
    28. ↵
      2. Savica R ,
      3. Carlin JM ,
      4. Grossardt BR , et al
      . Medical records documentation of constipation preceding Parkinson disease: A case-control study. Neurology 2009;73:1752–8. doi:10.1212/WNL.0b013e3181c34af5
      OpenUrlCrossRefPubMed
    29. ↵
      2. Suttrup I ,
      3. Warnecke T
      . Dysphagia in Parkinson’s disease. Dysphagia 2016;31:24–32. doi:10.1007/s00455-015-9671-9
      OpenUrlCrossRefPubMed
    30. ↵
      2. Polychronis S ,
      3. Dervenoulas G ,
      4. Yousaf T , et al
      . Dysphagia is associated with presynaptic dopaminergic dysfunction and greater non-motor symptom burden in early drug-naive Parkinson’s patients. PLoS One 2019;14:e0214352. doi:10.1371/journal.pone.0214352
    31. ↵
      2. Noyce AJ ,
      3. Silveira-Moriyama L ,
      4. Gilpin P , et al
      . Severe dysphagia as a presentation of Parkinson’s disease. Mov Disord 2012;27:457–8. doi:10.1002/mds.24006
      OpenUrlPubMed
    32. ↵
      2. Mu L ,
      3. Sobotka S ,
      4. Chen J , et al
      . Altered pharyngeal muscles in Parkinson disease. J Neuropathol Exp Neurol 2012;71:520–30. doi:10.1097/NEN.0b013e318258381b
      OpenUrlCrossRefPubMed
    33. ↵
      2. Mu L ,
      3. Sobotka S ,
      4. Chen J , et al
      . Alpha-synuclein pathology and axonal degeneration of the peripheral motor nerves Innervating pharyngeal muscles in Parkinson disease. J Neuropathol Exp Neurol 2013;72:119–29. doi:10.1097/NEN.0b013e3182801cde
      OpenUrlCrossRefPubMed
    34. ↵
      2. Masaoka T ,
      3. Tack J
      . Gastroparesis: Current concepts and management. Gut Liver 2009;3:166–73. doi:10.5009/gnl.2009.3.3.166
      OpenUrlCrossRefPubMedWeb of Science
    35. ↵
      2. Braak H ,
      3. de Vos RAI ,
      4. Bohl J , et al
      . Gastric alpha-synuclein immunoreactive inclusions in Meissner’s and Auerbach’s plexuses in cases staged for Parkinson’s disease-related brain pathology. Neurosci Lett 2006;396:67–72. doi:10.1016/j.neulet.2005.11.012
      OpenUrlCrossRefPubMedWeb of Science
    36. ↵
      2. Pasricha PJ ,
      3. Grover M ,
      4. Yates KP , et al
      . Functional dyspepsia and gastroparesis in tertiary care are interchangeable syndromes with common clinical and pathologic features. Gastroenterology 2021;160:2006–17. doi:10.1053/j.gastro.2021.01.230
      OpenUrlPubMed
    37. ↵
      2. Heetun ZS ,
      3. Quigley EMM
      . Gastroparesis and Parkinson’s disease: a systematic review. Parkinsonism Relat Disord 2012;18:433–40. doi:10.1016/j.parkreldis.2011.12.004
      OpenUrlPubMed
    38. ↵
      2. Zhang X ,
      3. Svn Z ,
      4. Liv M , et al
      . Association between irritable bowel syndrome and risk of Parkinson’s disease: A systematic review and meta-analysis. Front Neurol 2021;12:720958. doi:10.3389/fneur.2021.720958
      OpenUrl
    39. ↵
      2. Lai S-W ,
      3. Liao K-F ,
      4. Lin C-L , et al
      . Irritable bowel syndrome correlates with increased risk of Parkinson’s disease in Taiwan. Eur J Epidemiol 2014;29:57–62. doi:10.1007/s10654-014-9878-3
      OpenUrlCrossRefPubMed
    40. ↵
      2. Yoon SY ,
      3. Shin J ,
      4. Heo S-J , et al
      . Irritable bowel syndrome and subsequent risk of Parkinson’s disease: a nationwide population-based matched-cohort study. J Neurol 2022;269:1404–12. doi:10.1007/s00415-021-10688-2
      OpenUrl
    41. ↵
      2. Tornkvist NT ,
      3. Backman A-S ,
      4. Linder M , et al
      . Identification of irritable bowel syndrome in the Swedish National Patient Register: a validation study. Scand J Gastroenterol 2023;58:709–17. doi:10.1080/00365521.2023.2173021
      OpenUrl
    42. ↵
      2. Camilleri M ,
      3. Gorman H
      . Intestinal permeability and irritable bowel syndrome. Neurogastroenterol Motil 2007;19:545–52. doi:10.1111/j.1365-2982.2007.00925.x
      OpenUrlCrossRefPubMedWeb of Science
    43. ↵
      2. Skjærbæk C ,
      3. Knudsen K ,
      4. Horsager J , et al
      . Gastrointestinal dysfunction in Parkinson’s disease. J Clin Med 2021;10:493. doi:10.3390/jcm10030493
    44. ↵
      2. Drennan VM ,
      3. Rait G ,
      4. Cole L , et al
      . The prevalence of Incontinence in people with cognitive impairment or dementia living at home: a systematic review. Neurourol Urodyn 2013;32:314–24. doi:10.1002/nau.22333
      OpenUrlPubMed
    45. ↵
      2. Gabbard SL ,
      3. Lacy BE
      . Chronic intestinal pseudo-obstruction. Nutr Clin Pract 2013;28:307–16. doi:10.1177/0884533613485904
      OpenUrlCrossRefPubMed
    46. ↵
      2. Park S ,
      3. Kim J ,
      4. Chun J , et al
      . Patients with inflammatory bowel disease are at an increased risk of Parkinson’s disease: A South Korean nationwide population-based study. JCM 2019;8:1191. doi:10.3390/jcm8081191
      OpenUrl
    47. ↵
      2. Zhu F ,
      3. Li C ,
      4. Gong J , et al
      . The risk of Parkinson's disease in inflammatory bowel disease: A systematic review and meta-analysis. Dig Liver Dis 2019;51:38–42. doi:10.1016/j.dld.2018.09.017
      OpenUrl
    48. ↵
      2. Herrick MK ,
      3. Tansey MG
      . Is Lrrk2 the missing link between inflammatory bowel disease and Parkinson’s disease? NPJ Parkinsons Dis 2021;7:26. doi:10.1038/s41531-021-00170-1
    49. ↵
      2. Camacho-Soto A ,
      3. Searles Nielsen S ,
      4. Racette BA
      . Inflammatory bowel disease and risk of Parkinson's disease in Medicare beneficiaries. Parkinsonism Relat Disord 2018;57:77. doi:10.1016/j.parkreldis.2018.06.028
      OpenUrl
    50. ↵
      2. Yilmaz R ,
      3. Bayram E ,
      4. Ulukan Ç , et al
      . Appendectomy history is not related to Parkinson’s disease. J Parkinsons Dis 2017;7:347–52. doi:10.3233/JPD-171071
      OpenUrl

    Supplementary materials

    • Supplementary Data

      This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    Footnotes

    • Contributors BK: conceptualisation, methodology, formal analysis, data curation, validation, visualisation, writing-original draft, and writing-review and editing. LV: conceptualisation, resources, and investigation. JV: conceptualisation, methodology, and visualisation. GB: conceptualisation, resources, and investigation. RB: conceptualisation, resources, and investigation. MM: conceptualisation. KH: conceptualisation, methodology, and validation. GY: methodology and formal analysis. JT: writing-review and editing. PJP: guarantor, supervision, conceptualisation, methodology, validation, and writing-review and editing.

    • Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.

    • Competing interests None declared.

    • Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.

    • Provenance and peer review Not commissioned; externally peer reviewed.

    • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.