[1] 10w Clinical phenotype and risk of levodopa-induced dyskinesia in Parkinson's disease

[1] 251w It is unclear whether patients with different clinical phenotypes of Parkinson's disease (PD) differ in their risk of developing levodopa-induced dyskinesia. We evaluated the possible association between clinical phenotypes and risk of levodopa-induced dyskinesia in PD patients using a case-control design. The FRAGAMP study is a large Italian multicenter study. Patients affected by PD diagnosed according to the Gelb's criteria were enrolled and underwent a face-to-face interview. Clinical scales were used to evaluate motor and cognitive impairment. Presence of dyskinesia was assessed by the item 32 of the UPDRS section IV. On the basis of the most prominent motor symptoms at onset PD, patients were classified as tremor-dominant, akinetic-rigid, or mixed type. 485 PD patients (292 men; mean age 65.6 ± 9.8) were enrolled in the study of whom 128 (26.4 %) presented levodopa-induced dyskinesia. Of the 485 patients, 311 (64.1 %) were classified as tremor-dominant, 104 (21.4 %) as Akinetic-Rigid and 70 (14.4 %) as mixed type. Multivariate logistic regression analysis showed a significant negative association between tremor-dominant phenotype and levodopainduced dyskinesia (adjusted OR 0.48; 95 % CI 0.23-1.00; p value 0.05). When analysis was stratified by age at onset a stronger negative association was found among the late onset ([50 years) PD patients (OR 0.28; 95 % CI 0.11-0.70; p value 0.007) while no association was found among patients with an early onset. Our findings support the hypothesis that the occurrence of resting tremor as an initial manifestation of PD may predict a lower probability of developing levodopa-induced dyskinesia.

[1] 118w Parkinson's disease (PD) is a chronic progressive neurodegenerative disorder characterized by motor and nonmotor dysfunctions and levodopa is the most effective treatment currently available for the symptoms of PD [1]. Most patients experience motor response fluctuations within 5 years of starting treatment, and dyskinesias then frequently appear [2]. Although preclinical and clinical findings suggest pulsatile stimulation of striatal postsynaptic receptors as a key mechanism underlying levodopainduced dyskinesias (LID), their pathogenesis is still unclear [3]. Several factors have been associated with the development of LID such as disease duration, Hoehn-Yahr stage, average dosage and duration of levodopa treatment [4,5]. However, it is unclear whether patients with different clinical subtypes of PD differ in their risk of developing LID [6].

[2] 65w The FRAGAMP study (Fattori di Rischio Ambientali e Genetici Associati alla Malattia di Parkinson, that is Environmental and Genetic Factors in Parkinson's Disease) is a large multicenter case-control study carried out in Central-Southern Italy to evaluate the possible role of environmental and genetic factors in PD [7]. In the present study, we evaluated the possible relationship between clinical phenotype and the risk of developing LID.

[1] 65w At the end of the study period 585 PD patients, who fulfilled Gelb's criteria were enrolled [8]. Ninety-two (15.7 %) PD patients presented an adjusted MMSE \24 and were excluded from the analysis. One case was excluded from the analysis due to missing value. Seven of 492 PD cases without cognitive impairment were excluded from the final analysis because they were taking any antiparkinsonian treatment.

[2] 36w Finally 485 PD cases [292 men (60.2 %); mean age 65.6 ± 9.8 years] were included in the analysis of whom 128 (26.4 %) presented LID according to the item 32 of the UPDRS section IV.

[3] 36w Out of the 485 PD patients 439 (90.5 %) were taking levodopa [200 (41.2 %) levodopa alone and 239 (49.2 %) levodopa in combination with dopamine agonists] while 46 (9.5 %) were taking only dopamine agonists.

[4] 40w According to the symptom at onset of the 485 patients 311 (64.1 %) were classified as TD, 104 (21.4 %) as AR and 70 (14.4 %) as MT. Baseline characteristics according to the clinical phenotype are shown in Table 1.

[5] 75w At univariate analysis we found a positive association between presence of dyskinesia and age at onset, disease duration, Hoehn-Yahr stage and UPDRS-III score, average LED and duration of dopaminergic treatment. At univariate analysis, TD phenotype showed a reduced risk of developing LID respect to the AR patients. A negative association between body mass index (BMI) and coffee consumption and risk of dyskinesia was also found. Baseline characteristics and univariate analysis are reported on Table 2.

[6] 202w After adjustments on age, disease duration, Hoehn-Yahr stage, average LED intake, according to the LRT, the negative association between TD phenotype and LID was still significant with 52 % decreased risk of developing LID respect to the AR phenotype (adjusted OR 0.48; 95 % CI 0.23-1.00; p value 0.05) as shown in Table 3. When coffee consumption, protective factor for the risk of developing LID in our sample, was included in the multivariate model the strength of association between clinical phenotype and LID was not significantly changed (for TD the adjusted OR was 0.50; 95 % CI 0.23-1.06). It should be also noted that when gender was included in the multivariate model regardless the lack of association at the univariate analysis (OR 1.05; 95 % CI 0.69-1.58; p value 0.8) we found a significant positive association between female gender and LID with an OR of 1.91 (95 % CI 1.00-3.57; p value 0.05). However the inclusion of gender in the multivariate model has not significantly changed the strength of association between clinical phenotype and LID (for TD phenotype the OR was 0.47; 95 % CI 0.22-1.00; p value 0.05; for MT the OR was 0.59; 95 % CI 0.21-1.64; p value 0.3).

[7] 80w Disease duration and average LED were positively associated at the risk of dyskinesia at both univariate and multivariate analysis as shown in Tables 2 and 3. In particular in our sample the prevalence of dyskinesia was 2.8 % in the patients with a disease duration ranging from 0 to 3 years; 35.0 % in the group with a disease duration ranging from 4 to 6 years and 67.4 % among PD patients with a disease duration greater than 6 years.

[8] 278w Concerning the age at onset, out of the 485 PD patients 101 (20.5 %) had the onset of disease before the age of 50 years of whom 27 patients (26.7 %) had had the onset before the age of 40. TD phenotype was significantly more common among the late onset patients respect to the early onset (67.4 versus 51.5 %; p value 0.01), while AR was significantly more frequent among the early onset (33.7 versus 18.2 %; p value 0.01). Frequency of MT were similar in both early and late onset group (14.8 versus 14.3 %). Univariate analysis (Table 2) showed a three times increased risk of developing dyskinesia among subjects with an early onset (\50 years), even if such association was not longer observed at multivariate analysis (OR 0.65; 95 % CI 0.25-1.73; p value 0.4). Nevertheless when analysis was stratified considering the age at onset (early onset B50 years; late onset [50 years) the protective effect of TD phenotype was found just in the strata with an age at onset [50, while no association between clinical phenotype and LID was recorded among the early onset PD patients (\50 years). In particular, considering the AR phenotype as the reference group, in the group of late onset PD patients for the TD subtype the adjusted OR was 0.28 (95 % CI 0.11-0.70; p value 0.007) and for the MT the adjusted OR was 0.43 (95 % CI 0.13-1.43; p value 0.1). In the early onset group for the TD subtype the adjusted OR was 2.09 (95 % CI 0.51-8.60; p value 0.3) and for the MT the adjusted OR was 1.52 (95 % CI 0.19-11.7; p value 0.4).

[9] 17w Severity of dyskinesia evaluated by the AIMS was higher among AR phenotype respect to both TD (AIMS

[1] 276w PD is a heterogeneous disorder with variable clinicopathologic phenotypes and natural history. Several studies have confirmed that patients presenting with tremor at onset have a slower progression of disease than those with a AR subtype [16][17][18]. Even if different risk factors have been suggested to be associated with the occurrence of dyskinesia in PD patients [4,5], it is unclear whether patients with different clinical subtypes of PD differ in their risk of developing LID. To the best of our knowledge, only two studies have recently supported the hypothesis that resting tremor as an initial manifestation of PD predicts lower probability of developing LID [6,19]. In particular a first study was carried out in 2011 in a small cohort of 85 PD patients and, according to multivariate analysis, the absence of resting tremor as initial manifestations of PD was a strong predictor of LID with an OR of 8.67, independent of the duration of PD [19]. This finding was also confirmed by a larger cross-sectional study involving 367 PD participants in which the authors reported a negative association between TD phenotype and risk of LID with an OR of 0.52 [20]. We evaluated the risk of dyskinesia among TD and AR phenotypes and we have also found a 52 % decreased risk of developing dyskinesia among the TD subtype respect to the AR one, supporting the observation of a more benign course of PD in TD forms. In agreement with literature data and also in our sample, presence of LID was related to other well-recognized risk factors such as disease duration, Hoehn-Yahr stage and UPDRS-ME score, female gender, duration of dopaminergic drugs intake and average LED [4,5,20,21].

[2] 110w The idea that distinct PD subtypes involve different pathophysiological mechanisms may explain the reduced risk of LID in tremor-dominant patients. Tremor-dominant patients show different morphologic lesion patterns compared to bradykinesia and rigidity dominant patients. The tremor-dominant type shows more severe cell loss in the medial substantia nigra (SN), which projects to the dorsolateral striatum and ventromedial thalamus, causing hyperactivity of thalamo motor and cerebellar projections. For the akinetic-rigid type, a more severe cell loss occurs in the ventro lateral part of SN and posterior putamen, which causes inhibition of the glutamatergic thalamo-cortical pathway and reduced cortical activation. Different patterns of dopamine loss may reflect a variety of neuropathological features [22].

[3] 48w However, it should be noted that when analysis was stratified according to age at onset, a stronger and negative association was found among the PD patients with late onset while no association was found among the early onset patients (\50 years). Even if we have no clear [18,23].

[4] 163w Probably due to their low prevalence in the general population, few data concerning the clinical phenotype at onset of monogenic parkinsonisms are available, although tremor as initial sign has been reported in more than half patients with parkin mutations [24,25]. Thus, these genes, or some proteins related to the signalling pathways and functions of them, might be important for the pathogenesis of LID in early onset PD patients, independently by the clinical phenotype. Even if it is not clear, the underlying mechanism responsible of the increased risk of dyskinesia in early onset PD and monogenic parkinsonisms, recent neuroimaging studies have suggested a possible role of morphological and functional alterations involving the inferior frontal cortex and the supplementary motor area [26,27]. Furthermore, we cannot exclude the possible role of other genetic factors related to the risk of developing dyskinesia, such as the polymorphic status of the dopamine receptor D2 and catechol-O-methyltransferase genes as well as the Met allele of the brain-derived neurotrophic factor [28,29].

[5] 102w Out of the 101 PD patients classified as early onset (\50 years), 27 (26.7 %) presented an onset before the age of 40 and 50 % of these patients presented LID. It is possible to hypothesize that some of these patients can actually have a Parkin disease or other autosomal recessive forms of parkinsonism involving different neuropathological processes possibly also in the development of dyskinesia [30]. Nevertheless, we cannot exclude a possible random variation due to the small number of events in the early onset group (only 34 patients presented a TD subtype among the early onset group and 18 presented LID).

[6] 38w The size of our sample, the direct face-to-face interview to collect clinical, pharmacological and environmental data, the complete neurological examination including standardized clinical scales to evaluate both motor and cognitive impairment, represent the main strength of our study.

[7] 136w On the other hand, we are aware about the possible limits related to our study design that should be taken into account in interpreting our results. In particular the retrospective nature of the study does not allow us to exclude a possible lack of accuracy in the classification of the clinical phenotype based on the most prominent motor symptoms at onset, referred by the patients at the time of the enrolment. However, it should be underlined that our data are consistent with those reported by a previous community-based study, showing a frequency of 61 % tremordominant PD patients with an age at onset between 55 and 64 years [31]. Furthermore, even if we cannot exclude a possible misclassification of some patients, in this case a non-differential misclassification is expected reducing the risk of possible recall bias.

[8] 48w Another possible weakness is related to a possible selection bias due to the hospital-based design. However, characteristics of our PD sample are close to those reported in other population-based studies [7] and findings related to well recognized risk factors of LID are in agreement with literature data [4,5,20].

[9] 60w Our findings support the hypothesis that aside from the known clinical predictors of LID, occurrence of resting tremor as an initial manifestation of PD may predict not only slower progression of the disease [16][17][18], but also lower probability of developing LID. In-depth knowledge about these subtypes may lead to further insights into mechanisms of disease and pathogenesis-targeted and symptomatic treatments.

[10] 32w The study has been approved by the appropriate ethics committee and has been performed in accordance with the ethical standards laid down in the 1964 Declaration of Helsinki and its later amendments.

[11] 15w Informed consent All persons gave their informed consent prior to their inclusion in the study.

[1] 162w The FRAGAMP study is a large multicenter case-control study involving five movement disorder centers located in Central-Southern Italy. Patients affected by PD diagnosed according to the Gelb's diagnostic criteria [8], were consecutively enrolled in the study. All patients underwent a standard neurological examination performed by experienced neurologists in movement disorders, who had had a specific training before the beginning of the study. Hoehn-Yahr stage [9], unified Parkinson's disease rating scale (UPDRS) [10] were also used to evaluate PD patients. Furthermore the mini mental state examination (MMSE) [11] was administered to all the enrolled subjects in order to assess the presence of cognitive impairment and, to avoid the use of proxy responders and possible differences in the accuracy of recall, PD patients with a MMSE score lower than 24 were excluded from the analysis. Presence of LID was assessed according to the item 32 of the UPDRS section IV, while severity of dyskinesia was evaluated using the abnormal involuntary movement scale (AIMS) [12].

[2] 37w On the bases of the most prominent motor symptoms at onset of PD, patients were classified as tremor-dominant (TD), akinetic-rigid (AR), or mixed type (MT). Background and clinical characteristics of PD patients are extensively reported elsewhere [7].

[3] 82w PD patients enrolled in the study underwent a face-toface interview performed by trained neurologists. A standardised structured questionnaire made up of different sections, was administered to record demographic and clinical data, including pharmacological history, as well as data on environmental exposures. Data on cigarette smoking and alcohol consumption, both negatively associated with PD in the FRAGAMP study [13], were also collected. The study was approved by the local ethical committee and patients and controls were enrolled only after signed the informed consent.

[4] 75w Data were analyzed using STATA 12.1 software packages [14]. Data cleaning was performed before the data analysis considering both range and consistence checks. Quantitative variables were described using mean and standard deviation. The difference between means and the difference between proportions was evaluated by the t test and the Chi square test respectively. Linear relationship was explored by the Pearson's correlation coefficient (r). In case of not a normal distribution appropriate non-parametric tests were performed.

[5] 32w Presence of dyskinesia, the outcome in study, has been established on the bases of the item 32 of the UPDRS section IV, while the severity of dyskinesia was evaluated by the AIMS.

[6] 27w Unconditional logistic regression analysis was performed and for each study variable, we calculated OR, 95 % confidence interval (CI), and p value (two-tailed test, a = 0.05).

[7] 74w Multivariate analysis was performed to investigate the independent effect of a risk or protective factor after adjustment for one or several other factors or to adjust for confounding variables. Parameters associated with the outcome at the univariate analysis with a threshold of p = 0.10 were included in the model. The model was manually constructed using the likelihood ratio test (LRT) to compare the log-likelihood of the model with and without a specific variable.

[8] 24w Whenever variables were dichotomized or polychotomized, the cutoffs were derived from the pooled distribution of cases and control subjects (e.g., using the median value).

[9] 28w To evaluate the role of dopaminergic therapy the levodopa equivalent dose (LED) was calculated for those patients taking dopamine agonists or levodopa in combination with dopamine agonists [15].

[10] 50w The possible interaction was also evaluated by the LRT (test of violation of proportional odds). For quantitative exposure the test for linear trend was performed to evaluate the linear or trend effect. Stratified analyses were performed for PD onset (early onset defined as \50 years, late onset as [50 years).

[1] 19w Conflicts of interest On behalf of all authors, the corresponding author states that there is no conflict of interest.