Abstract
Background:
Both genetic and environmental factors contribute to Parkinson’s disease (PD) risk.
Objective:
We investigated the potential association of several relevant variables with PD age at onset (AAO), focusing on LRRK2 p.G2019S and GBA p.N370S mutations.
Methods:
Ashkenazi Jewish (AJ) PD patients, screened for LRRK2 and GBA mutations, underwent an interview regarding exposure to the following environmental and lifestyle factors: cigarette smoking, consumption of coffee, tea and alcohol, head injury and rural living. Multivariate linear regression (adjusted for sex) was used to examine the association with AAO, and models included LRRK2 p.G2019S and GBA p.N370S mutation status (carrier/non-carriers), single environmental variable and their interactions terms, as independent variables.
Results:
225 Israeli AJ PD patients were enrolled: 65 LRRK2 p.G2019S mutation carriers, 60 GBA p.N370S carriers and 100 non-carries of these mutations. In the dichotomized exposure/non-exposure analyses, positive LRRK2 p.G2019S status was associated with younger AAO in all models, at nominal or near significant levels (p = 0.033–0.082). Smoking was associated with older AAO (p = 0.032), and the interaction between GBA p.N370S and history of head injury was associated with younger AAO (p = 0.049), both at nominal significance. There was no indication of a consistent main effect for GBA p.N370S status or significant LRRK2 p.G2019S-environmental factor interaction. In the dose-dependent analyses, increased coffee and tea consumption levels were associated with older AAO (p = 0.001 and p = 0.002, respectively).
Conclusions:
Our results suggest that genetic and environmental factors may affect AAO in PD patients, but validation in additional samples is required.
Keywords
INTRODUCTION
Parkinson’s disease (PD) is the second most common neurodegenerative disease, caused by a multifactorial etiology that includes genetic susceptibility, exposure to environmental factors, aging processes and their interactions [1, 2]. Furthermore, a wide variability between PD patients is observed in age at onset (AAO), clinical manifestations and course of disease progression [3–5]. Genetic variation contributes to PD AAO, and heritability of this phenotype has been reported [6, 7]. PD risk loci have also been associated with disease AAO, some of them involved in alpha-synuclein mechanisms [7, 8].
Only a relatively small proportion of PD cases can be attributed to a single gene mutation [9, 10]. Among the Ashkenazi Jewish (AJ) population, two monogenic forms of PD are relatively common and well described: leucine-rich repeat kinase 2 (LRRK2) p.G2019S-related PD, and glucocerebrosidase (GBA)-related PD [11–14]. The LRRK2 p.G2019S mutation is found in 10% to 18% of sporadic AJ PD patients [11, 14] and in approximately 1% of sporadic PD patients worldwide [15]. Several GBA mutations associated with PD risk are found in nearly 20% of AJ patients, and the most frequent is p.N370S [11, 16–18]. Carriers of LRRK2 p.G2019S and at least some of the GBA mutations were reported to have younger AAO compared to non-carriers, but this finding is not consistent for LRRK2 p.G2019S [13–17, 19–22].
Several environmental and lifestyle factors have been associated with a decreased prevalence of PD and are therefore considered putatively protective [23]. Such factors include caffeine [24–29] and tea intake [28–30], a history of smoking [26, 32], ibuprofen use [33], alcohol consumption [26] and physical activity [34]. On the other hand, some factors increase the risk for PD, such as exposure to pesticides [26, 36], rural living and agricultural employment [26, 38], dairy consumption [39, 40] and a history of head injury [26, 42].
Most studies on environment and lifestyle factors in PD have focused on their role in disease risk. It seems plausible that at least few of these factors, which are associated with idiopathic PD risk or protection, influence (in a similar direction) disease AAO among affected individuals. For example, a positive smoking history (which is associated with reduced PD risk) was reported in respect to older AAO of motor signs among PD patients, while never-smokers had a younger AAO [43]. In parallel, we assume that risk factors for PD may be associated with younger AAO among exposed PD patients (e.g., rural living or previous head injury).
In the current study of Israeli AJ PD patients, we aimed to investigate the association of environmental and lifestyle factors with AAO among LRRK2 p.G2019S mutation carriers, GBA p.N370S carriers and non-carrier patients, and their interactions. Based on a “double-hit” model, AAO may be influenced by a combined contribution of mutation status and environmental exposure (younger AAO among mutation carriers who were also exposed to PD risk factors, or older AAO among non-carriers exposed to PD protective factors). To the best of our knowledge, the present study is the first to specifically address this question.
MATERIALS AND METHODS
The study is based on a sample of unrelated Israeli AJ patients with clinically diagnosed PD (according to the United Kingdom Brain Bank criteria) [44], regularly followed up at the Movement Disorders Institute at Sheba Medical Center. DNA samples of these patients (N = 679) were fully sequenced for LRRK2 and GBA genes, as described previously [12, 18]. In the present study, we included AJ patients who were either heterozygous for LRRK2 p.G2019S or GBA p.N370S mutations. We excluded patients with both mutations, carriers of any GBA mutation other than p.N370S, or homozygous for any of the two. Additional group (non-carriers) included 100 AJ PD patients negative for LRRK2 p.G2019S and any other GBA mutations, who were randomly selected from nearly 480 eligible individuals in the cohort.
The medical charts of the participants were reviewed by examiners blinded to the genetic status. Demographic data was retrieved, including year of birth and AAO. As a good proxy for AAO (and in similar approach to other studies [6, 7]), AAO was defined in this analysis as age at PD diagnosis (by a physician, according to medical records).
Participants were approached by telephone and offered to take part in the study. We were unable to approach 9 LRRK2 p.G2019S and 8 GBA p.N370S patients (some passed away since DNA collection). Consenting participants (most were aware of their genetic testing status) completed a short telephone interview, developed by the investigative team, about exposure to environmental factors and lifestyle habits. Demographic data was confirmed, and information was obtained regarding family history of PD (first- or second-degree relatives), cigarette smoking habits, consumption of coffee, tea (various types of caffeinated tea only) and alcoholic beverages (wine, beer and other types), history of head injury and rural living. In approximately 20% of the cases—usually for patients with cognitive impairment or language difficulties—information was completed with the assistance of a spouse or offspring familiar with the patients’ habits.
Various definitions of exposure to environmental factors have been used in PD studies [28, 42]. For the purposes of this study, smokers were defined as having smoked at least 100 cigarettes during their lifetime (consistent with the definition of ever smokers [45, 46]) prior to PD diagnosis. For coffee, tea and alcohol, individuals who reported regularly drinking at least one cup/glass per week for a minimum of a year before PD AAO were categorized as exposed (yes). For history of smoking, as well as coffee and tea consumption, participants reported the dosage (per day) and years of starting and quitting (if applicable), until AAO. Dosage and duration were determined, and exposure intensity (pack years for cigarette smoking, cup years for coffee/tea) was calculated (number of packs/cups per day, multiplied by years of consumption).
Participants were asked about head injury (with and without loss of consciousness) prior to PD diagnosis. Detailed residential history was recorded for each individual, and rural living was defined as living at least one year in a rural environment or location (village, farm or kibbutz) before PD AAO.
The study was approved by the Institutional Review Board of Sheba Medical Center and informed consent was obtained from all patients.
Statistical analysis
Frequencies and percentages were used for describing the categorical variables; means and standard deviations were used for continuous variables. Chi-square and Kruskal-Wallis tests were employed to study differences in demographic and background variables between LRRK2 p.G2019S mutation carriers, GBA p.N370S carriers and non-carriers.
The associations between LRRK2 p.G2019S mutation, GBA p.N370S mutation, environmental/lifestyle variables and AAO were analyzed by multivariate linear regression models. Each model included LRRK2 and GBA mutation status (carrier/non-carrier), a single environmental/lifestyle variable (cigarette smoking, coffee/tea/alcohol drinking, head injury and rural living) and their interaction terms, as independent variables for the prediction of AAO. Sex was included as a covariate in each model.
For all 6 environmental and lifestyle factors, analysis was performed as a dichotomized variable: exposed (yes) vs. not exposed. In order to test the interaction terms in the regression analysis, we employed centering [47,48, 47,48], and LRRK2 p.G2019S, GBA p.N370S as well as the environmental variables were mean-centered prior to the analyses.
In addition, a dose-dependent analysis of total exposure level was performed for smoking, coffee and tea (only the exposed individuals were included in this analysis). Due to the retrospective design of the study, potential inaccuracies in exposure level quantification were expected, and therefore these continuous measures were divided into tertiles (low, medium and high) to minimize this. These 3 models were analyzed in a similar way to the dichotomized models described above.
All statistical tests were two-sided, and p-value ≤0.05 was defined as statistically significant (nominal level). Applying a correction for multiple testing would have required p≤0.00185 (0.05/27, taking into account 9 statistical models, each included environmental factor and two mutations). All analyses were performed using IBM-SPSS Statistics version 25 (IBM corp., Armonk, NY, USA).
RESULTS
A total of 225 AJ PD patients were included in the final analysis (136 males, mean AAO of 60.0±9.7 years). As shown in Table 1, 65 (28.9%) were LRRK2 p.G2019S mutation carriers (39 males, mean AAO 58.6±10.4 years), 60 (26.7%) GBA p.N370S carriers (34 males, mean AAO 59.0±9.4 years) and 100 (44.4%) were negative for both mutations (63 males, mean AAO 61.5±9.2 years). Mean PD AAO among non-carriers patients was older, but the difference was not statistically significant (p = 0.105). As expected, we observed a significant correlation between PD age at diagnosis (defined as AAO in this study) and age at first motor manifestation reported by the patient (r = 0.976, p < 0.001). AAO data by mutation status, environmental factors and lifestyle habits is provided in Supplementary Table 1.
Comparing demographic, environmental and lifestyle variables frequencies among the three groups, we found that positive family history was significantly more common among LRRK2 p.G2019S carriers (56.9%) compared to GBA p.N370S carriers and non-carrier patients (30.0% and 23.0%, respectively) (p < 0.001). Head injury frequency was higher in LRRK2 p.G2019S mutation (24.6%) and GBA p.N370S (33.3%) carriers compared to 13.0% in non-carriers (p = 0.009), while rural living was more frequent in non-carriers (51.0%), compared to 38.5% and 25.0% for LRRK2 p.G2019S and GBA p.N370S carriers, respectively (p = 0.005, Table 1).
Demographic, environmental and lifestyle characteristics of the study sample. Significant results are in bold. Unless otherwise specified, data refer to number (N) and percentage
SD – standard deviation; PD – Parkinson’s disease, AAO – age at onset.
The inter-correlations between the six environmental and lifestyle variables (exposed/not exposed) were small (r < 0.10) and insignificant (p > 0.05) apart from the correlation between coffee and alcohol consumption, which was significant (p = 0.014), but with a small effect size (r = 0.164).
Table 2 presents the regression models testing the association between PD AAO and status of LRRK2 p.G2019S and GBA p.N370S mutations (carrier/non-carriers), exposure (yes/no) to environmental and lifestyle variables, and their interactions.
Summary of multivariate linear regression analysis models for PD AAO prediction. Each model included a different environmental/lifestyle factor (dichotomized variable – exposed versus not exposed). Significant associations (p < 0.05) are in bold
CI – confidence interval.
Adjusting for sex, positive LRRK2 p.G2019S mutation status was nominally associated with younger PD AAO in the head injury model (β= –0.16, p = 0.033) and the rural living model (β= –0.15, p = 0.038) and was at near significant level (0.06 < p < 0.082) in all other models of cigarette smoking, coffee, tea and alcohol consumption. Positive GBA p.N370S status was nominally associated with younger AAO in the rural living model (β= –0.16, p = 0.036), but did not reach significance in the other models (although the effect was consistent with the expected direction).
With regard to environmental and lifestyle variables, smoking was nominally associated with AAO, indicating older PD AAO among patients who smoked (β= 0.14, p = 0.032). A near significant association between rural living and AAO (β= –0.13, p = 0.068) was observed (younger AAO among PD patients who lived in rural places). No other variables were significant.
Examining gene-environment interactions, a nominally significant association was found for the interaction between GBA p.N370S and head injury (β= –0.16, p = 0.049). Thus, among mutation carriers, a younger AAO was detected in those who experienced head injury. There was no other GBA or LRRK2 p.G2019S-environment interactions in all models examined (Table 2).
Including only the subgroup of exposed individuals, a dose-dependent analysis of exposure level to smoking, coffee and tea (divided into tertiles) is presented in Table 3. Consumption levels of coffee (β= 0.24, p = 0.001) and tea (β= 0.29, p = 0.002) were significantly associated with PD AAO (indicating older AAO among patients who reported drinking larger amounts of these beverages). The association of coffee withstood our threshold for multiple testing correction (p≤0.00185), but for tea it was only at a nominal significance. Positive GBA p.N370S status was nominally associated with younger AAO in the coffee model (β= –0.17, p = 0.029), while positive LRRK2 p.G2019S status in none of the models.
Summary of multivariate linear regression analysis models for PD AAO prediction. Each model studied the intensity of the exposure (expressed as tertiles - low, medium and high) to a different variable. Significant associations (p < 0.05) are in bold
CI – confidence interval.
A retrospective power analysis for the mutations, based on a ratio of ∼2.5 between non-carriers and carriers (specifically: 2.46 for LRRK2 p.G2019S and 2.75 for GBA p.N370S), found that the sample sizes required for achieving a small effect size (Cohen’s d = 0.2) under α= 0.05 and 1-β= 0.80 are 543 and 217 (non-carriers and carries, respectively), while for a medium effect (d = 0.5), the required sample sizes are 89 and 35. Our sample included approximately 160 non-carriers and 60 carriers (of each mutation).
With regards to the smoking status variable, based on a ratio of ∼1.3 between non-smokers and smokers, the required samples for achieving a small effect size (Cohen’s d = 0.2) under α= 0.05 and 1-β= 0.80 are 357 and 275 (non-smokers and smokers, respectively); for a medium effect (d = 0.5) the required sample sizes are 59 and 45. Our sample included 127 non-smokers and 98 smokers.
Importantly, the reported associations (except for coffee exposure level) were not statistically significant after correcting for multiple testing. Therefore, these results should be considered preliminary, and require replication in additional samples.
DISCUSSION
While most studies of environmental factors and lifestyle habits in the context of PD have focused on their impact on disease risk, this study examined the effect of several relevant variables on AAO among PD patients who are carriers or non-carriers of LRRK2 p.G2019S and GBA p.N370S mutations. Our sample offers a valuable opportunity to investigate the interplay between genetic and environmental factors that may contribute to AAO, in a relatively homogenous population of AJ PD patients living in Israel, screened for the most frequent PD-causing mutations in this origin.
Overall, following a dichotomous definition (yes/no) of exposure to environmental and lifestyle variables and their inclusion in the regression models, PD AAO among patients with LRRK2 p.G2019S mutation was younger at nominal or near significant levels in all models (in agreement in some, but not all studies). We found a similar direction of effect among carriers of p.N370S (considered a mild GBA mutation [16]), but overall it did not reach the required level of statistical significance. Since several other studies have reported a younger AAO of PD among carriers of GBA mutations [13, 22], we hypothesize that in a larger sample these results would have been significant.
Positive smoking history was nominally associated with older AAO, in the regression model that included LRRK2 p.G2019S and GBA p.N3720 S mutation status. However, we did not find an interaction between smoking and the mutations. In addition, a lack of dose-dependent association was observed between smoking intensity and AAO among the smokers, although the direction of the effect was consistent.
Nicotine has a neuroprotective effect on neurons, including dopaminergic ones [49], and it modulates the release of dopamine by these cells [50]. In a large meta-analysis, smoking history was associated with reduced PD risk by approximately 36%, and the effect was significant among all types of smoking status categories (current vs. never, ever vs. never, past vs. never) [26]. Another meta-analysis found that compared with never smokers, PD relative risk was 0.59 for ever smokers, 0.80 for past smokers, and 0.39 for current smokers [27]. Among PD patients, in line with its neuroprotective role, smoking was associated with older AAO [43, 51]. A possible involvement of genetic factors (for example, variation of nicotinic receptor genes) was suggested to modulate the relationship between smoking and PD AAO [52]. However, another potential explanation for the link between PD and smoking is reverse causality, due to the loss of responsiveness to nicotine reward and changes in nicotinic receptors [53].
The association of rural living with PD AAO approached significance level (younger AAO among those who lived in rural areas, as expected). This is in agreement with reports of rural living as a risk factor for PD [26, 37], although other studies were less consistent [31]. Notably, we did not assess the contribution of other rural living related factors, such as farming or exposure to pesticides (by themselves risk factors for PD [26, 38]), and only some of the participants who lived in rural areas were practicing agriculture. Therefore, the effect of rural living on PD AAO should be further studied in the future.
On the other hand, we observed a dose-dependent (continuous data divided into tertiles) association of coffee or tea consumption with AAO. Among regular drinkers of these beverages, a higher consumption level was associated with older AAO. Interestingly, the dichotomized variables for coffee and tea were not significantly associated with AAO. At least for coffee, this may be explained by the small group of non-drinkers, which provided this analysis with less statistical power.
Various studies have demonstrated that caffeine have a protective effect on PD development [24–29]. This link may be mediated by adenosine receptor signaling, since caffeine is an adenosine A2A receptor antagonist [54, 55]. The inverse association of tea consumption with PD risk and AAO is also known (but inconsistent), and some have suggested that this effect is specific to black tea [28–30, 51].
Interestingly, we did not find a significant association of alcohol consumption or previous head injury with AAO. Despite the lack of a significant main effect, the interaction between GBA p.N370S and head injury was nominally significant: mutation carriers with a history of head injury had younger AAO, in line with the “double-hit” model. However, this should be interpreted cautiously since the sample size within each subgroup was small, and the definition of positive head injury history (with or without loss of consciousness) was based on participants’ self-report and not on medical records, leading to potential biases and inaccuracies.
The major advantage of the current study is the availability of relatively large groups of LRRK2 p.G2019S and GBA p.N370S carriers, with demographic, clinical, environmental and lifestyle data. All participants were AJ in a long-term follow up at a single, tertiary movement disorders referral clinic. The majority of them have lived most of their adult life period in Israel, and therefore may share some degree of background exposure level to environmental factors (although some immigrated to Israel at a relatively advanced age). Moreover, the study design included a control group of mutation-negative AJ patients (non-carriers), thus enabling to differentiate the contribution of environmental exposure to AAO from that of the two mutations. Inconsistencies with results from previous studies regarding factors influencing PD AAO could be attributed to a lack of adjustment for relevant genetic and environmental variables.
Several limitations of our analysis should be acknowledged, including the retrospective design, based on self-reported data from an interview administrated at later life, and therefore possiblly leading to various biases. In approximately 20% of the cases, data was completed by interviewing family members (rather than the patients themselves), and this may have added to the inaccuracies (mainly regarding the patients’ early life information). Moreover, we did not use a previously standardized questionnaire about exposure habits and quantities. In order to mitigate the effect of inaccurate consumption data, we used tertiles of exposure level for the dose-dependent analyses (rather than continuous variables). Taken together, the study design was not optimal, and should ideally be performed prospectively.
The small sample size (N = 225) limits the statistical power to detect significant associations. Except for coffee consumption dose, all findings were significant at a mere nominal level, and did not withstand correction for multiple testing. A larger sample size would have been an advantageous, possibly strengthening the results by allowing them to withstand the adjustment for multiple testing, and to find association of GBA p.N370S and/or other environmental factors with AAO, including interactions. Nevertheless, the convergence of the expected and observed effect directions for the two mutations and some of the environmental and lifestyle factors reduces the likelihood of false positive results. Last, several environmental factors which are known to affect PD risk were not examined in this study, such as exposure to pesticides and other toxins, or use of non-steroidal anti-inflammation drugs (NSAIDs).
In conclusion, it is of interest to further investigate the contribution of both genetic and environmental factors to AAO of PD, as well as to other clinical features, such as rate of disease progression. Due to the discussed limitations, we regard the results of this study as preliminary at the current stage. Future follow up studies in larger samples are required, preferably in a prospective design, in order to reach more definitive conclusions.
CONFLICT OF INTEREST
Dr. Gilad Yahalom and Dr. Simon Israeli-Korn received consultancy fees from Abbvie Biopharmaceuticals Inc. Dr. Ziv Gan-Or received consultation fees from Idorsia, Inception Sciences, Denali, Prevail Therapeutics and Lysosomal Therapeutics Inc. Prof. Sharon Hassin-Baer received consultancy fees from Abbvie Biopharmaceuticals Inc., Boston Scientific, Teva Pharmaceuticals, Medtronic and Actelion LTD. No other disclosures were reported.
Footnotes
ACKNOWLEDGMENT
Dr. Ziv Gan-Or holds the Fonds de recherche du Québec – Santé (FRQS) Chercheurs-Boursiers Junior 1 award, granted by FRQS and Parkinson Quebec, and the New Investigator Award granted by Parkinson Canada.
