Cardiac autonomic function during exercise and incident Parkinson’s disease

Introduction

Parkinson’s disease (PD) is the world’s fastest growing neurological disorder in terms of prevalence¹. One main clinical challenge lies in earlier detection, as significant neurological degeneration has already occurred by the time PD is identified with the prevailing diagnostic approaches². At present, prediction models for the early diagnosis of PD consist of predominantly non-modifiable, binary clinical characteristics (e.g. male gender, or presence of hyposmia), resulting in rather ‘static’ models^3,4. In contrast, a more dynamic variable that could reflect quantitative changes over time, with repeated measurements during follow-up, would offer additional insights. Such a biomarker could improve early diagnosis and capture the gradual progression towards manifest PD^2,5. In this context, electrocardiographic (ECG) parameters that focus on cardiac autonomic measures have been proposed as a potentially important tool^6,7.

In patients with manifest PD, altered cardiac autonomic function is common^8,9 including impaired heart rate regulation at rest and during exercise¹⁰. Accumulating evidence suggests that dysregulation of autonomic processes in general, and cardiac autonomic dysfunction in particular, may precede the motor symptoms of PD^6–8. Importantly, pathological studies in PD have demonstrated a characteristic sequence of neurodegenerative events, with early involvement of the dorsal motor nucleus of the vagal nerve¹¹, which regulates (cardiac) parasympathetic tone. Cardiac imaging studies have also demonstrated changes in the sympathetic nervous system in very early stages of PD¹². This background has motivated different initiatives in population-based cohorts to search for potential prodromal autonomic markers, including heart rate regulation¹³.

Owing to differences in population characteristics, variations in choice of autonomic markers, and non- uniformity in protocols used to assess these markers, the available evidence is difficult to interpret^14–17. Two large cohort studies focused on heart rate variability (HRV), a marker of cardiac parasympathetic tone, as a potential prodromal marker for incident PD and reported conflicting findings^14,15. Based on differences in maximum heart rates achieved, a small controlled exercise study suggested impaired sympathetic activity as prodromal sign for PD¹⁶, but this could not be confirmed in a larger cohort¹⁷. Here, we focus on other autonomic markers, namely the increase in heart rate during exercise and its reduction during recovery, the advantage being that these markers are easily extractable, straightforward to assess, and clinically established in different settings¹⁸.

We therefore studied these markers in a long-term follow-up study of 69,288 UK Biobank participants who performed an exercise test, which represents the largest investigation to date on cardiac autonomic parameters and incident PD. Our specific aim was to assess whether exercise-related parameters of cardiac autonomic function during a baseline bicycle test would be associated with incident PD during long-term follow-up.

Methods

Participants

The UK Biobank study comprises a total of 502,364 participants, with even numbers of men and women aged 40-69 years at recruitment from 21 assessment centers across England, Wales, and Scotland, with extensive baseline and follow-up clinical, biochemical, genetic and outcome measures. The study received approval from the North West Multi-Centre Research Ethics Committee, and all participants provided informed consent at the time of enrolment (2006 – 2010). From the full cohort, 96,524 participants consented to participate in an exercise stress test with heart-rate monitoring (2009-2013). According to protocol, 13,962 participants were considered ineligible to perform the activity (https://biobank.ndph.ox.ac.uk/ukb/ukb/docs/Cardio.pdf). Another 7221 participants who did not complete the test were excluded as well. Furthermore, exclusions were based on a history of cardiovascular disease (N=4441), a diagnosis of cancer within 1 year before or after the exercise stress test (N=1526) and a history of a neurodegenerative (N=260) disorder. In particular, we focused on the exclusion of patients with PD at the time of the exercise test (N=97) using previously described diagnostic criteria (https://biobank.ctsu.ox.ac.uk/crystal/ukb/docs/alg_outcome_pdp.pdf). Thus, self-reported diagnoses were also considered, to ensure that all potential cases with PD were excluded at baseline.

In total, 69,288 of the 96,524 (71.8%) participants who provided consent to participate in the exercise test were included in the present analysis (Figure 1).

• Download figure
• Open in new tab

Figure 1.

Study exclusion diagram The full list of UK Biobank variables used for disease exclusions is provided in Supplemental Table 1. ^a Participants ineligible to exercise had one of more safety risk factors as defined in paragraph 9.1 of the protocol manual provided by UK Biobank (https://biobank.ndph.ox.ac.uk/showcase/ukb/docs/Cardio.pdf). ^b Excluded subjects who met one or more of the listed criteria; CV, cardiovascular; PD, Parkinson’s disease.

Results

In total, there were 96,524 participants who consented to partake in the exercise test, of whom 27,236 were excluded (Figure 1, Supplemental Table 2). Excluded subjects showed clinically relevant differences in baseline demographics (higher age and BMI) and in established prodromal factors of incident PD (higher prevalence of risk factors) compared to the remaining 69,288 individuals (Supplemental Table 3).

Incident PD was diagnosed in 319 individuals (0.5%) (Table 1). At the time of the exercise test, participants who developed PD during follow-up were older than those who remained disease free and were more often male.

Rates of many of the prodromal risk factors (such as constipation) were higher, while the rate of smoking (protective factor) was lower.

Median follow-up duration on survival status was 12·5 years (IQR: 0·3) with more than 10 years of follow-up available in 59,460 participants (85.6%). During this follow-up period, 3106 (4.5%) participants died.

Heart rate response to recovery is blunted in prodromal PD

During the pre-test stage, resting heart rate was not significantly different between participants who developed PD and those who did not (Table 2). However, the increase in heart rate during exercise (HRI-exc) and decrease during recovery (HRD-rec) were both lower in the group that developed PD (Table 2).

View this table:

• View inline
• View popup
• Download powerpoint

Table 2 Heart-rate markers during exercise testing comparing participants with and without incident PD during long-term follow-up

Results of the univariate cox regression analysis are presented in (Supplemental Table 4). The established prodromal clinical variables were significantly associated with incident PD. With regard to cardiac autonomic parameters, HRI-exc was associated with a 30% higher risk of incident PD per 10 beats less increase in heart rate during exercise. For HRD-rec, a blunted lowering in heart rate from peak exercise to recovery was associated with a 60% higher risk of incident PD per 10 beats difference.

In the minimally adjusted cox regression analysis, HRD-rec was significantly associated with incident PD, and HRI-exc was not (Table 3). After adjusting for clinical and autonomic prodromal factors, it was found that a blunted lowering in heart rate during recovery remained independently associated with a 30% higher risk of incident PD: HR 1.3; 95% CI 1.1-1.4; p < 0·001 per 10 beats less recovery (Table 3, Supplemental Table 5).

View this table:

• View inline
• View popup
• Download powerpoint

Table 3 Association of exercise heart-rate markers with incident PD, sequential adjustment for clinical and autonomic factors

Figure 2 displays time-to-event curves. The median lag time to (hospital) diagnosis of incident PD was 9.3 years (IQR: 4.4.), showing the highest risk of incident PD in the subjects with the lowest HRD-red. After adjusting for clinical and autonomic prodromal factors, the association between HRD-rec and incident PD was found to be approximately log-linear (Figure 3).

• Download figure
• Open in new tab

Figure 2.

Unadjusted Kaplan-Meier estimates stratified by heart rate decrease during recovery (HRD-rec)

• Download figure
• Open in new tab

Figure 3 Association of heart rate decrease during recovery (HRD-rec) with risk of Parkinson’s disease.

Adjusted for sex, age, ethnicity, Townsend deprivation index, body mass index, type-2 diabetes, smoking status, bladder dysfunction, constipation, depression, and self-reported sleep duration and sleeplessness; the number above each vertical line is the hazard ratio, and the number below is the number of incident cases.

Discussion

The present report with 319 new cases of PD represents the largest population-based cohort to date on cardiac parameters of autonomic function as a potential prodromal marker, assessed by changes in heart rate during exercise and recovery. We found that an impaired capacity to reduce the heart rate during the recovery phase immediately after exercise was associated with incident PD, independently of established prodromal clinical and autonomic markers. Notably, our findings do not indicate a threshold value, but rather suggest a more graded association between a blunted heart rate recovery after exercise and incident PD (Figures 2 and 3). There was a univariate association between heart rate increase during the exercise phase and incident PD, which was no longer present after adjustment for sex and age. We also replicated earlier work by showing that subjects with prodromal parkinsonism had a greater likelihood of having established prodromal markers at baseline, including signs and symptoms indicative of non-cardiac autonomic dysfunction (i.e. constipation etc.). This suggests that the cohort studied here was representative of other prodromal populations described in the literature. Analogous to other established prodromal markers²³, the lag time to diagnosis was long, with a median time of 9 years. This time interval represents the time until diagnosis made during an in-patient stay, so the actual lag time to a diagnosis established on an outpatient basis is shorter. Notably, while most prodromal markers are ‘static’ binary and non-modifiable variables (i.e. sex, hyposmia), we now present a candidate marker that is continuous and well known to be modifiable by interventions. Therefore, this marker has potential to be ‘dynamic’ as it may reflect quantitative changes over time by use of repeated measurements during follow-up. Longitudinal studies are warranted. As a quantifiable, continuous parameter, this cardiac autonomic marker may not only offer early diagnostic value, but also has promise to capture more subtle progressive changes during the prodromal phase. If proven to be modifiable in PD, it might be relevant when used as outcome measure in studies of putative disease-modifying interventions applied in the prodromal phase^24–26.

In contrast to heart rate variability, that was studied in previous cohorts ^14,15, heart rate recovery is a highly reproducible measure of parasympathetic activity^18,20, with a straightforward protocol to register and calculate the parameter, using data contained within a standard exercise test. Heart rate recovery is also a recognized indicator of cardiovascular fitness²⁷. The fact that participants who developed PD during follow-up versus those who did not show marked differences in other autonomic prodromal markers at the time of the exercise test (constipation, bladder dysfunction, etc) underscores that heart rate recovery in this subset of participants is a marker of altered autonomic function. This may well fit into the concept that the dorsal nucleus of the vagal nerve is one of the first manifestations in the neuropathological sequalae that lead towards clinically manifest motor PD^8,11,28. Moreover, the association between HRD-rec and incident PD was similar in direction and magnitude when participants with a lower and higher workload were studied separately (Supplemental Table 6). With regard to previous cohort studies, the large Atherosclerosis Risk in Community (ARIC) cohort studied heart rate variability at rest and found an association with incident PD. During 20 years of follow-up in 12 162 patients (57% women, mean age 54 years), PD was diagnosed in 0·6% (78 cases)¹⁴. The smaller Cardiovascular Health Study (n= 5888) studied a community-based population of individuals who were 65 years and older (58% women, mean age 73 years). In this study heart rate variability was assessed during 24-hour Holter monitoring¹⁵. During 14 years of follow-up, PD was diagnosed in 3% of individuals, and in the subset (n=1587) with Holter monitoring no association was found between HRV and incident PD¹⁵. Although experimental studies have shown that HRV is sensitive to systemic cholinergic blockade²⁹, the interpretation of HRV measures remains considerably more complicated than heart rate recovery, as it is strongly dependent on other physiological factors, including heart rate, and depth and frequency of respiration^29,30. This aspect, together with non- uniformity between studies in the assessment of HRV (for example, 2 min resting ECG versus 24-hour registrations), hampers the interpretation of currently available evidence on HRV.

The increase in heart rate during exercise is primarily driven by sympathetic activation^18,19. Both in patients with manifest motor PD, and in individuals in the prodromal phase, sympathetic involvement has been demonstrated^8,12,28,31. A small case-control study on heart rate profiles during maximal exercise showed that the 18 studied cases had lower heart rates at peak exercise than the 36 matched controls, at a mean of 4 years prior to diagnosis. Heart rate recovery was not studied. A similar, but larger study did not detect significant signs of sympathetic dysfunction during the premotor phase of PD¹⁷.

In the present study, we observed that the heart rate increase during exercise was lower for individuals who developed PD during follow-up, but the association was no longer present after adjustment for sex and age, suggesting that sympathetic dysfunction is not an independent marker. Explanations for the discrepant findings with the small physiological study may lie in methodological aspects such as maximal versus submaximal exercise³², differences in sample size, or the selection of the incorporated confounding variables.

Our study is not without shortcomings. First, we acknowledge limitations in terms of generalizability with this rather healthy study cohort. Yet, with this particular study population, the association between heart rate recovery and incident PD is less likely to be flawed by co-morbidity. Second, the present work is a post-hoc analysis on a database where incident PD was not the primary outcome of interest. We therefore applied a strict case definition, that can be supported by source documents (a hospital admission confirming the ICD-code for PD). Notably, the hospital-based diagnosis was systematically collected during follow-up, in contrast to out-of- hospital diagnoses. The marked differences in baseline (prodromal) profiles between incident PD cases and the remainder of the group speaks for the solidity of our case definition, and the clinical representativeness of the cohort. Potentially, this case definition may have caused an underestimation of the actual incidence and may have influenced the effect estimations. Moreover, such a definition may result in a different case load in that patients with a hospital diagnosis may vary in clinical profile from those with an out-of-hospital diagnosis.

This is why we performed a post-hoc sensitivity analysis in the subset of participants with both information on the out-of-hospital and in-hospital phase using an alternative definition for incident PD, i.e. new out-of-hospital and/or in-hospital diagnosis of PD (Supplemental Table 7). The incidence of new cases was 0.5%, and results with regard to the association between HRD-rec and the new outcome measure were in line with the overall conclusions of the main analysis. With regard to the clinical profile of incident PD cases, we observed that in the aforementioned subset of participants, about 30% had an earlier out-of-hospital diagnosis. The diagnosis was made 2.4 years earlier than the in-hospital diagnosis. Appreciating that the majority of in-hospital diagnoses were first diagnoses, an interesting hypothesis is that these may have been more severe cases, as early involvement of body structures outside the nervous system (including the gut and the heart) is associated with a more malignant phenotype and a more rapid disease progression¹².

We also acknowledge that the differential diagnosis from atypical parkinsonism (where the nature and extent of autonomic dysfunction is different) can be challenging, and a misclassification rate of up to 15% occurs in clinical studies of PD patients. However, such a misclassification would not have markedly affected the associations that we found, because early and prominent problems with cardiac innervation are not expected in most forms of atypical parkinsonism. Finally, our registrations did not meet the required duration and physical conditions for assessment of heart rate variability²⁹, which unfortunately precludes comparisons with the previous cohorts^14,15.

We foresee several areas for further research. First, our findings remain to be confirmed to provide the scientific background for initiatives to study its potential predictive value. To allow for a pooled comparison across studies that tested the same autonomic parameter, it is imperative to focus on better standardization of assessment protocols^29,30. Second, with the growing notion of distinct patterns in the development towards PD, defining a brain-first and a body-first subtype¹², more detailed studies are warranted in these subpopulations^24,25. It is plausible that the potential diagnostic value will be greater in the body-first subtype, where the autonomic nervous system is involved much earlier in the disease process.

Taken together, our findings provide additional supportive evidence for the concept that pathophysiological changes in the autonomic nervous system may precede the clinically manifest motor syndrome of PD. If confirmed in an independent cohort, we anticipate that autonomic cardiac biomarkers could be incorporated in a more comprehensive battery of early diagnostic biomarkers^5,33. Potentially, as a quantitative, potentially modifiable marker, it may pave the way for longitudinal biomarker studies on therapeutic interventions.

Author Contributions

S.vD, M.B., B.B., A.D., contributed to the conception and design of the study, S.vD, M.B., J.R., M.O., J.I., B.B., A.D. contributed to the acquisition and analysis of data; S.vD, M.B., J.R. J.S. S.D, M.O., A.T., P.M. J.T., L.E., J.I., P.L., B.B, and A.D. contributed to drafting the text and preparing the figures.

Potential conflicts of interest

None of the authors has any conflict of interest report that is relevant to the present publication. Outside the present work, BRB has the following to report. He is co-Editor in Chief for the Journal of Parkinson’s disease. He is on the editorial board of Practical Neurology and Digital Biomarkers, has received honoraria from being on the scientific advisory board for Abbvie, Biogen, and UCB, has received fees for speaking at conferences from AbbVie, Zambon, Roche, GE Healthcare, and Bial, and has received research support from the Netherlands Organization for Scientific Research, the Michael J Fox Foundation, UCB, Not Impossible, the Hersenstichting Nederland, the Parkinson’s Foundation, Verily Life Sciences, Horizon 2020, and the Parkinson Vereniging (all paid to the institute). AD’s research team is supported by a range of grants from the Wellcome Trust [223100/Z/21/Z, 227093/Z/23/Z], Novo Nordisk, Swiss Re, Boehringer Ingelheim, National Institutes of Health’s Oxford Cambridge Scholars Program, EPSRC Centre for Doctoral Training in Health Data Science (EP/S02428X/1), and the British Heart Foundation Centre of Research Excellence (grant number RE/18/3/34214). JR acknowledges fellowship RYC2021-031413-I and grant PID2021-128972OA-I00, both funded by MCIN/AEI/10.13039/501100011033 and the European Union ‘‘NextGenerationEU/PRTR’’. PDL is supported by UCL/UCLH Biomedicine NIHR. This work acknowledges the support of the National Institute for Health and Care Research Barts Biomedical Research Centre (NIHR203330); a delivery partnership of Barts Health NHS Trust, Queen Mary University of London, St George’s University Hospitals NHS Foundation Trust and St George’s University of London. PBM and AT acknowledge the support of the National Institute for Health and Care Research Barts Biomedical Research Centre (NIHR203330); a delivery partnership of Barts Health NHS Trust, Queen Mary University of London, St George’s University Hospitals NHS Foundation Trust and St George’s University of London, and Medical Research Council grant MR/N025083/1.

Data availability statement

Data from the UK Biobank are available to researchers on application to the UK Biobank (https://www.ukbiobank.ac.uk/enable-your-research/about-our-data).