Abstract
Background:
Parkinson’s disease (PD) subjects are less likely to ever smoke and are more prone to quit smoking, as compared to controls. Therefore, smoking habits can be considered part of the non-motor phenotype, preceding the onset of motor PD by several years.
Objective:
To explore non-motor symptom (NMS) correlates of smoking habits in de novo PD.
Methods:
This cross-sectional study included 281 newly diagnosed, drug-naïve PD subjects, recruited in Naples (Italy) and in Kassel (Germany). All subjects completed the NMS Questionnaire (NMSQ), and were investigated for smoking status (never, current and former smokers) and intensity (pack-years).
Results:
140 PD subjects never smoked, 20 currently smoked, and 121 had quit smoking before PD diagnosis. NMSQ total score did not associate with smoking status, but with smoking intensity (p = 0.028; coefficient = 0.088). A multinomial logistic regression stepwise model presenting never smoking as reference, selected as NMSQ correlates of current smoking: sex difficulties (p = 0.002; OR = 5.254), daytime sleepiness (p = 0.046; OR = 0.085), insomnia (p = 0.025; OR = 0.135), and vivid dreams (p = 0.040; OR = 3.110); and of former smoking: swallowing (p = 0.013; OR = 0.311), nausea (p = 0.027; OR = 7.157), unexplained pains (p = 0.002; OR = 3.409), forgetfulness (p = 0.005; OR = 2.592), sex interest (p = 0.007; OR = 0.221), sex difficulties (p = 0.038; OR = 4.215), and daytime sleepiness (p = 0.05; OR = 0.372). An ordinal logistic regression stepwise model selected as NMSQ correlates of smoking intensity: nocturnal restlessness (p = 0.027; coefficient = 0.974), and leg swelling (p = 0.004; coefficient = 1.305).
Conclusions:
Certain NMSs are associated with different smoking status and intensity, suggesting a variety of adaptive mechanisms to cigarette smoking.
INTRODUCTION
There is significant epidemiological evidence supporting an inverse relationship between smoking and risk of Parkinson’s disease (PD) [1]. In particular,several studies found that PD subjects are less likely to ever establish a smoking habit [2, 3]. In this view, smoking has been suggested to protect against PD, by interacting with nicotine receptors and with other receptor-independent mechanisms involved in oxidative stress, neuroprotection or in neuroinflammation [4–6]. Alternatively, PD patients have been reported to be more prone to quit smoking as compared to controls [2, 7], hence suggesting that the ease of smoking cessation could be considered part of the early non-motor symptom (NMS) or premotor profile of PD [2, 8], possibly presenting also a genetic background [9]. In fact, NMSs can precede the onset of motor PD by several years, arguably as a consequence of a generalized neurodegenerative process starting in the dorsal brainstem and olfactory bulb, and gradually developing throughout the brain [1, 11].
With particular regard to the association between PD and smoking, an early loss of cerebral nicotinic receptors might be responsible for reduced sensitivity of the brain reward system in response to nicotine, with subsequent modifications of smoking habits. However, the interaction of these pathological changes with smoking exposure might also affect cognitive and behavioral aspects of PD [2, 12]. In line with this, lower sensation seeking behaviors have been associated with a reduced likelihood of ever smoking in PD [12], and, in addition, cognitive functions are apparently worse among PD smokers [13]. Therefore, it is also possible the association of different smoking habits with a general NMS profile [8]. In view of this, the present study aims to explore the extent to which different NMS are more likely associated with smoking habits in a population of newly diagnosed, drug naïve PD individuals. As secondary objectives, smoking habits have been studied in relation to motor variables and to activities of daily living.
MATERIALS AND METHODS
Study design and population
This is a cross-sectional study performed to evaluate non-motor correlates of different smoking habits in newly diagnosed, drug naïve PD subjects.
PD subjects have been retrospectively included from two different populations recruited in specialized centres for diagnosis and treatment of PD, at the “Federico II” University Hospital in Naples (Italy), and at the Paracelsus-ElenaKlinik in Kassel (Germany). Local ethical committees approved the studies and all subjects provided written informed consent. The study was performed in accordance with good clinical practices and the Declaration of Helsinki.
Inclusion and exclusion criteria have been fully reported elsewhere [8, 14]. Briefly, inclusion criteria were: new diagnosis of PD according to the United Kingdom Parkinson’s Disease Society Brain Bank Diagnostic Criteria [15, 16]; de novo criteria for dopaminergic treatments (no previous exposure, or no longer than 2 weeks and not within 4 weeks prior to study entry). Exclusion criteria were: symptoms or signs of secondary or atypical parkinsonism [16–20]; treatment with antipsychotic or with any other drug possibly affecting clinical evaluation.
Clinical evaluation
Demographics and clinical data were recorded. Disease duration (duration of first reported motor symptoms till diagnosis) was recorded. All subjects completed the NMS Questionnaire (NMSQ), a validated tool for detection of NMS [21, 22]. The NMSQ consists of 30 questions with dichotomous (yes/no) answers and of a total score, which ranges between 0 and 30, with higher scores reflecting more NMS [21, 22]. Motor disability was evaluated with the Unified Parkinson’s Disease Rating Scale (UPDRS) part III. Activities of daily living were evaluated with the UPDRS part II. Subjects were categorized according to the prevalent motor phenotype: postural instability/gait difficulty (PIGD), tremor dominant (TD), or indeterminate [23].
Smoking habits were recorded. PD subjects were categorized according to self-reported smoking status as never, former, or current smoker [8, 14]. In addition, current and former smokers were asked about their age at starting and the number of cigarettes smoked per day. Former smokers were also asked about the age at quitting.
Statistics
Means and proportions of demographics (age, gender, nationality) and clinical features (UPDRS part II, UPDRS part III, NMSQ) were calculated for the entire PD population, and for smoking status subgroups (current, former, and never smokers). Then, differences in demographics and clinical features in relation to smoking status were evaluated with t-test, Mann-Whitney test, χ 2 test, Fisher’s exact test, or analysis of variance (ANOVA), as appropriate.
Considering the main objective of the present study (smoking-related NMS profile), smoking habits were selected as outcome measures. NMSQ total score, and its single items were selected as main variables of interest. Covariates included in the analyses were age, gender and disease duration. Associations between non-motor variables (NMSQ total score) and smoking status (never, former and current smoking) were explored by using multinomial logistic regression, subsequently adjusted for age, gender and disease duration. Then, NMSQ items were included in a multinomial logistic regression stepwise analysis with backward selection for p = 0.20 as the critical value for entering variables in the model, while the smoking status was the dependent variable. Items selected by the stepwise analysis were subsequently included in a final model including age, gender and disease duration. Odds ratio (OR) and 95% confidence intervals (95% CI) were calculated. Associations between UPDRS part II total score, UPDRS part III total score and motor subtypes, and smoking status were explored by using multinomial logistic regression, subsequently adjusted for age, gender and disease duration.
Ever smokers (current and former smokers) were evaluated for pack-years (packs of cigarettes per day multiplied by years smoked), an estimate of the lifelong smoking intensity, and were categorized accordingly (smoking intensity subgroups: 0–9, 10–19, 20–29, 30–39, ≥40 pack-years) [1, 3]. Associations between different clinical features and pack-years subgroups were explored by using ordinal logistic regression analysis, subsequently adjusted for age, gender, disease duration and smoking status (current or former smoker). NMSQ items were included in an ordinal logistic regression stepwise analysis with backward selection for p = 0.20 as the critical value for entering variables in the model, while pack-years subgroups were dependent variables. Items selected by the stepwise analysis were subsequently included in a final model including age, gender and disease duration.
Covariates used for each model were tested for multicollinearity, considering values of multicollinearity diagnostics (VIF) less than 2.5 as an assumption of reasonable independence among covariates. Variables were tested for normal distribution by using both statistical and graphical methods when appropriate. Results have been considered statistically significant if p < 0.05. Stata 12.0 and Microsoft Excel have been used for data processing andanalysis.
RESULTS
Demographics, clinical features and smoking habits of the population are reported in Table 1. In the PD population, 140 subjects have never smoked (49.8%), 20 subjects were current smokers (7.1%), and 121 subjects (43.1%) quit smoking before PD diagnosis. Smoking subgroups were not different for age (p = 0.623), and disease duration (p = 0.333). Males were more frequently current or former smokers (p < 0.001) (Table 1). Smoking status (p = 0.854), and intensity (p = 0.512) were not different in relation to nationality (Italian or German).
Smoking status
NMSQ total score did not associate with smoking status, with similar likelihood of NMS presence for current smokers (p = 0.338; and p = 0.227 after adjusting for age, gender and disease duration) and for former smokers (p = 0.371; and p = 0.334 after adjusting for age, gender and disease duration), as compared to never smokers (Fig. 1A). However, following a multinomial logistic regression stepwise model presenting never smoking as reference, the presence of following NMSQ items was more likely associated with current smoking: sex difficulties (p = 0.002), daytime sleepiness (p = 0.046), insomnia (p = 0.025), and vivid dreams (p = 0.040) (Table 2). Whereas, the presence of following NMSQ items was more likely associated with former smoking: swallowing (p = 0.013), nausea (p = 0.027), unexplained pains (p = 0.002), forgetfulness (p = 0.005), sex interest (p = 0.007), sex difficulties (p = 0.038), and daytime sleepiness (p = 0.050) (Table 2) (Fig. 2).
Lower UPDRS part III total scores were more likely associated with current smoking (p = 0.015; and p = 0.042 after adjusting for age, gender and disease duration), and to former smoking (p = 0.004; and p = 0.032 after adjusting for age, gender and disease duration), as compared to never smoking (Fig. 1B). Considering motor subtypes, TD subjects, were more likely to be current smokers (p = 0.028; OR = 5.479; 95% CI = 1.206–24.887; and p = 0.027; OR = 5.634; 95% CI = 1.215–26.117 after adjusting for age, gender and disease duration), but not former smokers (p = 0.470; OR = 0.824; 95% CI = 0.489–1.390; and p = 0.465; OR = 0.814; 95% CI = 0.470–1.411 after adjusting for age, gender and disease duration), as compared to never smokers.
Lower UPDRS part II total scores were more likely associated with current smokers (p = 0.011; and p = 0.046 after adjusting for age, gender and disease duration) and to former smokers (p = 0.022; and p = 0.048 after adjusting for age, gender and disease duration), as compared to never smokers (Fig. 1C).
Smoking intensity
Current smokers presented longer smoking duration (p < 0.001), and higher smoking intensity (p < 0.001), as compared to former smokers (Table 1). Age did not associate with smoking intensity subgroups before (p = 0.875), and after adjusting for gender, disease duration and smoking status (p = 0.793).
NMSQ total score associated with smoking intensity subgroups before (p = 0.045), and after adjusting for age, gender, disease duration and smoking status (p = 0.028) (Fig. 3). Moreover, an ordinal logistic regression stepwise model selected the presence of following NMSQ items to be associated with smoking intensity: nocturia (p = 0.056), nocturnal restlessness (p = 0.027), leg swelling (p = 0.004) (Table 2).
UPDRS part III total score did not associate with smoking intensity subgroups (p = 0.533; coefficient = 0.012; 95% CI = –0.025–0.049; and p = 0.282; coefficient = 0.021; 95% CI = –0.017–0.061 for the model adjusted for age, gender, disease duration and smoking status). Motor subtype did not associate with smoking intensity (p = 0.729; coefficient = 0.122; 95% CI = –0.570–0.814; and p = 0.854; coefficient = 0.066; 95% CI = –0.643–0.776 for the model adjusted for age, gender, disease duration and smoking status).
UPDRS part II total score did not associate with smoking intensity subgroups (p = 0.758; coefficient = 0.013; 95% CI = –0.070–0.096; and p = 0.420; coefficient = 0.035; 95% CI = –0.051–0.123 after adjusting for age, gender, disease duration and smoking status).
DISCUSSION
The present study investigated possible relationships between NMSs and smoking habits, showing that the NMS profile at PD diagnosis might be differently associated with smoking status and intensity. Considering that chronic exposure to cigarette smoking results in a variety of adaptive mechanisms [24–26], it is possible that smoking habits are associated with the heterogeneous non-motor profile of PD [27].
To be more precise, current and former smokers more often complained of memory difficulties, which have been deemed a predictor of mild cognitive impairment [28]. Accordingly, current smokers at PD onset have previously been reported to have worse cognitive progression over time, than never smokers [13]. As a possible explanation, individuals may find smoking reinforcing because of nicotine-related cognitive facilitation [29], and this effect may become especially prominent among those presenting cognitive disorders [29, 30]. Therefore, a similar effect cannot be excluded for pre-motor PD subjects more at risk of cognitive impairment. Alternatively, it is also possible that cognitive disturbances are directly related to chronic cerebral damage, as a consequence of smoking [29, 30].
Moreover, depressive symptoms were more likely to be reported by current smokers, and significantly increased with smoking intensity. Interestingly, cigarette smoking seems to be both cause and effect of depressive symptoms in the general population [31], whereas sustained smoking cessation can improve emotional symptoms [32]. From a pathophysiological perspective, constituents of tobacco smoke might contribute to increased sensitivity in developing and maintaining behavioral symptoms [33], as previously suggested for impulse control disorders in PD smokers [34]. However, it has to be reported that a previous study failed to find any significant effect of transdermal nicotine patches on depressive symptoms of PD, evaluated with the Hamilton Depression Scale [35], and additional studies should be conducted to further explore this issue.
In addition, current smokers complained of fewer sleep disturbances, with a significant effect of smoking intensity [36]. Intriguingly, in the general population, smokers are more likely to have problems staying asleep than non-smokers, possibly because of nicotine withdrawal occurring during the night, but also because of concomitant respiratory issues that were unfortunately not investigated in the current study [37]. Hence, the interpretation of these findings is not straightforward. One possibility would be that PD patients experience less nocturnal nicotine withdrawal symptoms since, described in general, PD have reduced rewards from nicotine assumption [37]. Nonetheless, there was an increased likelihood of intense dreaming among current smokers. In particular, vivid dreams can be included in the clinical profile of REM sleep behaviour disorder (RBD) [38], and smoking is commonly reported in both idiopathic and PD-related RBD [39, 40].
Furthermore, nocturnal restlessness has been associated with increased smoking intensity, as previously shown for restless legs syndrome [41]. Interestingly, a similar effect was also present for leg swelling, although it is difficult to demonstrate whether this symptom represents a different descriptive term for restless legs symptoms [42], or a sign of impaired vascular reactivity due to smoking in combination with PD [10, 43].
Moreover, sex difficulties were more likely reported by ever smokers, as expected since this is a well-known side effect of cigarette smoking [44]. Similarly, gastrointestinal symptoms were differently reported among never, current and former smokers, possibly because of nicotine effects on the gastrointestinal tract, inducing, for instance, nausea [24].
Finally, there are also interesting results concerning motor variables, such as reduced motor scores in ever smokers at PD diagnosis, as compared to never smokers, suggesting a possible symptomatic effect. Worth of note, it has previously been showed that chronic nicotine exposure improves motor skills in a fly model of Parkin-disease [45]. Similarly, several clinical trials have been conducted to assess feasibility and efficacy of nicotine on PD motor symptoms, but definite results are yet to come [4]. However, a previous study using transdermal patches in PD suggested that the symptomatic effect of nicotine might be unnoticed because of concomitant anti-parkinsonian drugs and, thus, present results on drug-naïve PD patients might have a particular relevance [35]. Moreover, TD subjects were more likely to currently smoke, as previously suggested in a similar study [46]. It is worth noting that TD subjects have been previously described to present less NMS, as compared to PIGD [46]. In this view, since quitting smoking has been considered part of early non-motor features of PD [2, 8], the increased prevalence of TD smokers is possibly due to the reduced prevalence of NMSs in TD subjects at PD onset [46]. Furthermore, daily activities were less likely to be impaired among ever smokers, as compared to never smokers, in line with different motor disability.
However, there are limitations to be reported, such as the use of the NMSQ instead of the NMS scale or of specific neuropsychiatric or cognitive evaluations that unfortunately were not available in the entire population, although the NMSQ has previously demonstrated to be a useful clinical tool for screening non-motor features of PD in a clinical setting [47]. Moreover, information regarding passive smoking or different comorbidities possibly also affecting behavioural aspects of PD, was not recorded for the entire population. In addition, considering that smoking is an uncommon feature of PD and that larger on-going studies presenting full NMS evaluation in de novo populations did not include smoking habits [48], we tried to analyse a sample as large as possible, by retrospectively selecting PD subjects included in two different on-going studies with similar recruiting criteria [8, 14]. Moreover, the presence of a control group would have strengthen current results, although there are several previous studies on smoking effects in the general population that have been discussed. Finally, the inclusion of de novo PD subjects can be biased by misdiagnoses as for atypical Parkinsonisms, which is expected to be high, especially in the first years after the diagnosis [49], but allows evaluating NMSs and smoking habits before any possible confounding factor from disease course and pharmacological treatments.
In conclusion, although there are several methodological and clinical issues deserving further investigations, the present study showed for the first time that specific motor symptoms and heterogeneity of non-motor features in de novo PD are associated with smoking habits. Therefore, the evaluation of the NMS profile in relation to cigarette smoking seems particularly promising for an appropriate phenotyping of PD and, possibly, for future treatment approaches.
CONFLICT OF INTEREST
Authors declare that there are no conflicts of interest with regard to the current work.
Marcello Moccia received grants from the Federico II University, (Naples, Italy).
Brit Mollenhauer received independent research grants from TEVA-Pharma, Desitin, Boehringer Ingelheim, GE Healthcare and honoraria for consultancy from Bayer Schering Pharma AG, Roche, AbbVie, TEVA-Pharma, for presentations from GlaxoSmithKline, Orion Pharma, TEVA-Pharma and travel costs from TEVA-Pharma.; is member of the executive steering committee of the Parkinson Progression Marker Initiative of the Michael J. Fox Foundation for Parkinson’s Research and has received grants from the BMBF, EU, Deutsche Parkinson Vereinigung, Michael J. Fox Foundation for Parkinson’s Research, Stifterverband für die deutsche Wissenschaft, and has scientific collaborations with Roche, Ely Lilly, Covance and Biogen Idec; is listed as co-inventors in a patent application to the US Patent Office related to the quantification of α-synuclein in biological fluids for the purpose of improved diagnosis.
Roberto Erro received grants form the Univesity of Verona (Verona, Italy).
Marina Picillo received salary from the Federico II University (Naples, Italy), from the University of Salerno (Salerno, Italy), and from the Division of Neurology, Toronto Western Hospital (Toronto, Ontario, Canada); and grants from the Michael j Fox Foundation for Parkinson’s research.
Raffaele Palladino received salary from the Imperial College (London, United Kingdom), and grants from the Federico II University, (Naples, Italy).
Paolo Barone received salary from the University of Salerno (Salerno, Italy).
Footnotes
ACKNOWLEDGMENTS
The present study received no specific economic support.
