Abstract
Objective
Dyskinesia is a known side-effect of the treatment of Parkinson's Disease (PD). We examined the influence of haplotypes in three dopamine receptors (DRD1, DRD2 and DRD3) and the Brain Derived Neurotrophic Factor (BDNF) on dyskinesia.
Methods
Patient data were drawn from a population-based case-control study. We included 418 patients with confirmed diagnoses by movement disorder specialists, using levodopa and a minimum three years disease duration at the time of assessment. Applying Haploview and Phase, we created haploblocks for DRD1-3 and BDNF. Risk scores for DRD2 and DRD3 were generated. We calculated risk ratios using Poisson regression with robust error variance.
Results
There was no difference in dyskinesia prevalence among carriers of various haplotypes in DRD1. However, one haplotype in each DRD2 haploblocks was associated with a 29 to 50% increase in dyskinesia risk. For each unit increase in risk score, we observed a 16% increase in dyskinesia risk for DRD2 (95%CI: 1.05-1.29) and a 17% (95%CI: 0.99-1.40) increase for DRD3. The BDNF haploblock was not associated, but the minor allele of the rs6265 SNP was associated with dyskinesia (adjusted RR 1.31 (95%CI: 1.01-1.70)).
Conclusion
Carriers of DRD2 risk haplotypes and possibly the BDNF variants rs6265 and DRD3 haplotypes, were at increased risk of dyskinesia, suggesting that these genes may be involved in dyskinesia related pathomechanisms. PD patients with these genetic variants might be prime candidates for treatments aiming to prevent or delay the onset of dyskinesia.
Keywords: Parkinson's disease, dyskinesia, DRD1-3 and BDNF genes, haplotypes
Introduction
Parkinson's disease (PD) leads to significant disability and loss of quality-of-life.[1] One important component to loss of quality-of-life is L-dopa-induced dyskinesia, a common side-effect of treatment. Dyskinesia affects 25% within the first five years [2] increasing up to 80% among patients ten years into disease.[3] Dyskinesia is associated with depression and increases health-related costs.[4] Interestingly, although some patients develop dyskinesia early in their disease, other patients never do.
The two basal ganglia circuitry pathways (D1 and D2) associated with movement control are regulated by dopamine receptors. Common genetic variations in DRD loci are natural candidates for dyskinesia risk.[5,6] In addition, Brain Derived Neurotrophic Factor (BDNF) may affect dyskinesia due to modulation of dopamine receptor expression.[7,8]
The majority of studies focused on DRD1-3 and BDNF. Many of these have concentrated on SNPs only, specifically the rs1800497 in DRD2,[9–17] the rs6280 in DRD3,[11,13,18] and the rs6265 in BDNF.[9,19,20] The results from these studies have reported inconsistent results.[9–20] One small clinic-based study using a Brazilian population, analyzed DRD2 haplotypes (based on rs228365, rs1076560, rs6277, rs1800497 and rs2734849) and implicated that DRD2 haplotypes are associated with dyskinesia.[14]
Here we are using a targeted approach for the three dopamine receptors (DRD1-3) and BDNF, to estimate risk of dyskinesia based on haplotypes' variants. Reviewing haplotypes instead of SNPs limits the number of tests and allows us to examine gene regions. Patients were enrolled in our large population-based study in California (N=747) and dyskinesia was assessed relatively early, on average five years after first diagnosis based on cardinal motor symptoms.
Methods
Study population
We assembled a cohort from the population-based case control study Parkinson's Environment and Gene (PEG) study, which enrolled and followed patients from three Central California counties (Kern, Fresno, and Tulare) between 2001 and 2015. All patients were seen by movement disorder neurologists (JB, Dr. Bordelon) at least once at baseline, many on multiple follow-up occasions, and were confirmed as having probable idiopathic PD according to published criteria.[21] Recruitment occurred in two phases. Recruitment during the first phase (PEG1) in 2001-2007 has been described before.[22] Among 563 potential patients, 359 incident PD patients within the first three years after diagnosis were identified at baseline. Dyskinesia was assessed during the first follow-up appointment, where we saw 250 PEG1 patients at least once. The second recruitment strategy (PEG2) ensued during 2010-2015 and identified 388 idiopathic PD patients (from 589 potential patients). In PEG2, patients were allowed to have received a PD diagnosis after 2001.
Disease duration was measured from the date of diagnosis to the time of the Unified Parkinson's disease Rating Scale (UPDRS) IV assessment. We restricted our population to those taking levodopa (N=459; 193 PEG1 and 266 PEG2 subjects). As dyskinesia is unlikely to develop within the first years after diagnosis, we further restricted our analysis to patients seen with a minimum of three calendar years of disease duration. Among patients who were seen too early in disease (N=101) or were not taking levodopa medication (N=143) at the first assessment, we examined whether they met our criteria during a second assessment (34 PEG1; 26 PEG2). Supplemental figure e-1 provides a flow diagram for the study population of 418 PD patients included in our final analyses.
All study protocols regarding human subjects in PEG1 and PEG2 have been approved by the local Institutional Review Board and all participants gave their written informed consent.
Definition of outcome and other variables
Dyskinesia was assessed during follow-up visits in PEG1, and starting with the first visit in PEG2 (for some patients dyskinesia was assessed shortly after their first contact) using the UPDRS part IV.[23] Dyskinesia presence was noted (yes/no) and the severity of dyskinesia was measured in hours per day (no dyskinesia, ≤25%, 26-50%, 51-75% and >75% of the waking day).
Based on Parkinson's medication at the dyskinesia assessment, we calculated a levodopa-equivalent daily dosage (LEDD) converting all reported anti-Parkinsonian medications into a standardized equivalent dosage according to a previously described algorithm.[24] Race/ethnicity was established via genetic Ancestry Informative Markers (AIMS) if available (72.5% of those included here) or by self-report of ancestry if AIMS markers had not been measured (27.5%). We grouped race/ethnicity as European ancestry, Hispanic and other.
Genetic analysis
DNA was extracted at UCLA from whole blood and genotyping was conducted at UBC for all patients with a confirmed diagnosis of PD. A total of 26 SNP were selected to be genotyped across the three DRD genes: 5 SNPs for DRD1, 12 SNPS for DRD2, 8 for DRD3 loci (supplemental table e-1). Markers were mainly selected for haplotype tagging. Genotyping was performed in duplicate at UBC using a Sequenom MassARRAY platform and custom iPLEX assay. Three SNPs in the BDNF locus were previously genotyped at UCLA for a subset of the study sample (N=312 patients with PD). All SNPs had a genotyping call rate >97%, MAF>5% and were in Hardy-Weinberg equilibrium (as defined by a p-value greater than 0.01 in Caucasians) except for rs11030104 (p-value: 1.7*10-9), which was subsequently excluded.
Using Haploview,[25] we created linkage disequilibrium (LD) plots based on the total study population (supplemental figure e-2). A haploblock, based on several SNPs, consists of various haplotypes, i.e. SNP patterns. Haploblocks were created based on the confidence intervals.[26] The haplotypes with the highest probabilities for each subject were generated with Phase v2.1.1; rare haplotypes with a frequency of less than 30 patients (=3.7%) were combined into one subgroup.[27] We created one haploblock for the BDNF- and DRD1, and three haploblocks for DRD2 and DRD3 loci (supplemental figure e-2). The haplotypes in the three haploblocks are correlated. For the DRD2 haploblocks, the D' between the first and second haploblock is 1.0, and between the second and third 0.25. For DRD3, the D' between the first and second haploblock is 0.94 and between the second and third haploblock 0.61. The BDNF haplotype consisted of only two SNPs; the R-squared was 0.19 and D-prime was 1.0.
Statistical methods
All analyses were performed using SAS 9.4 (SAS Institute, Cary NC). We estimated risk ratios (RR) using Poisson regression models with robust error variance and a log link function[28] to estimate effects for eight chromosomal regions in four genes on dyskinesia occurrence. The analyses were adjusted for gender, PEG recruitment period, disease duration, LEDD, race/ethnicity and age at diagnosis. Throughout, we adjust for LEDD, adjusting for L-DOPA daily dosage instead produced the same results. For analyses of dyskinesia severity, we distinguished three categories [no, mild (<25%) and moderate/severe (>25%)] to ensure that the proportional odds assumption of our model was met. When multiple haploblocks in one gene showed associations, we created a risk score summing across the three risk-haplotypes and estimated the effect according to the number of risk haplotypes a patient carried.
Finally, we assessed whether individual SNPs alone could explain the associations by analyzing each SNP independently. In sensitivity analyses, we examined whether results were consistent across race/ethnicity by restricting participants to their ancestry.
Results
Patient characteristics
The prevalence of dyskinesia in our patient sample was 24.6%, with the majority experiencing symptoms of dyskinesia less than 25% of their waking hours per day. The average age at diagnosis was 67 years, and on average 5.5 calendar years had passed since first diagnosis when we assessed dyskinesia in our study (see table 1). Haplotype frequencies ranged between 4 and 51% in our haploblocks (supplemental figure e-2). The BDNF haplotype had a minimum frequency of 18.9% and a maximum of 45.4%. Because haploblock 2 of the DRD3 loci consists almost exclusively of the GG haplotype, we lacked statistical power to examine associations for this block. This haploblock was therefore omitted from analyses.
Table 1. Overview of characteristics of the PEG study population.
| Dyskinesia | No Dyskinesia | Total sample population | ||||
|---|---|---|---|---|---|---|
| Total N | N=103 % | Total N | N=315 % | Total N | N=418 % | |
| Race | ||||||
| European ancestry | 87 | 84.5 | 244 | 77.5 | 331 | 79.2 |
| Hispanic | 14 | 13.6 | 60 | 19.0 | 74 | 17.7 |
| Other | 2 | 1.9 | 11 | 3.5 | 13 | 3.1 |
|
| ||||||
| Smoker | ||||||
| Non-smoker | 58 | 56.3 | 171 | 54.3 | 229 | 54.8 |
| Former smoker | 43 | 41.7 | 138 | 43.8 | 181 | 43.3 |
| Current smoker | 2 | 1.9 | 6 | 1.9 | 8 | 1.9 |
|
| ||||||
| Gender | ||||||
| Male | 62 | 60.2 | 207 | 65.7 | 269 | 64.4 |
| Female | 41 | 39.8 | 108 | 34.3 | 149 | 35.6 |
|
| ||||||
| Family history of PD | ||||||
| No | 87 | 84.5 | 259 | 83.5 | 346 | 83.8 |
| Yes | 16 | 15.5 | 51 | 16.5 | 67 | 16.2 |
|
| ||||||
| Dyskinesia in hours | ||||||
| no dyskinesia | 0 | 0.0 | 315 | 100.0 | 315 | 75.4 |
| < 25% of waking day | 70 | 68.0 | 0 | 0.0 | 70 | 16.7 |
| >26 of waking day | 33 | 32.0 | 0 | 0.0 | 33 | 7.9 |
|
| ||||||
| Mean | SD | Mean | SD | Mean | SD | |
|
| ||||||
| Years of education | 14.2 | 4.2 | 14.0 | 4.7 | 14.1 | 4.5 |
|
| ||||||
| PD duration at time of measuring dyskinesia | 6.2 | 2.3 | 5.3 | 2.1 | 5.5 | 2.1 |
|
| ||||||
| Levodopa equivalent daily dosage | 739.3 | 488.9 | 601.4 | 340.4 | 635.4 | 386.3 |
|
| ||||||
| Age of diagnosis PD | 64.1 | 10.9 | 67.6 | 10.0 | 66.7 | 10.3 |
|
| ||||||
| UPDRS score | 26.7 | 9.9 | 24.0 | 12.0 | 24.7 | 11.6 |
N: number. %: percentage. SD: Standard deviation
Dyskinesia risk decreased with increasing age of PD diagnosis (with each year increase in age at diagnosis: RR=0.98 (95% CI: 0.96-0.99), increased with LEDD (per 1000 mg: RR=1.77 (95% CI: 1.29-2.44), increased with disease duration (per calendar year: RR=1.14 (95%CI: 1.07-1.21); and increased with UPDRS score (per five units increase: RR=1.07 (95%CI: 1.01-1.14)). Gender, years of schooling, smoking status, family history of PD, and ethnicity were not associated with dyskinesia in our population sample. The effect sizes and confidence intervals were very similar for the association among these variables with ‘severity of dyskinesia’.
Association haplotypes with dyskinesia and ‘severity of dyskinesia’
We did not find any association between dyskinesia or ‘severity of dyskinesia’ and haplotypes for DRD1. For DRD2, in both adjusted and unadjusted analyses, we found all three DRD2-haploblocks to be associated with dyskinesia. We adjusted for gender, PEG1 or 2 recruitment, disease duration, LEDD, race/ethnicity and age at diagnosis. In each haploblock there was one haplotype that accounted for about a 29 to 50% increased risk for dyskinesia. For ‘severity of dyskinesia’, the effect sizes were slightly higher (33-62% increase) and 95% confidence intervals (CIs) were narrower. For DRD3, the TGG-haplotype of the first haploblock and the CC-haplotype of the third haploblock suggested an increase in risk of developing dyskinesia, however the 95% CIs included the null value. Finally, for BDNF, the TA-haplotype was associated with ‘severity of dyskinesia’ with an increase of 43%, however the effect estimate for presence of dyskinesia did not reach statistical significance in our data (see tables 2 and 3).
Table 2.
Poisson regression analyses for the association between dyskinesia and the haplotypes for the DRD1, DRD2, DRD3 and the BDNF loci in PD patients (N=418 patients) with a minimum of 3 years PD disease duration.
| N | Crude | Adj | |||
|---|---|---|---|---|---|
| Dysk / No Dysk | RR | RR | 95%CI | p-value | |
| DRD1 (rs4867798|rs5326|rs265981) | |||||
| TCA | 67 / 212 | Ref | |||
| TCG | 64 / 211 | 0.97 | 1.00 | 0.75 – 1.34 | 0.98 |
| CCG | 34 / 109 | 1.01 | 1.06 | 0.74 – 1.52 | 0.75 |
| CTG | 30 / 87 | 1.07 | 1.14 | 0.78 – 1.66 | 0.50 |
| DRD2, block 1 (rs10891549|rs1554929|rs1124493|rs2242592|rs2245805) | |||||
| CTAGG | 78 / 293 | Ref | |||
| TCGTT | 60 / 183 | 1.17 | 1.20 | 0.90 – 1.60 | 0.22 |
| TCAGG | 48 / 103 | 1.51 | 1.50 | 1.09 – 2.07 | 0.01 |
| TCGTG | 5 / 30 | 0.68 | 0.63 | 0.26 – 1.52 | 0.30 |
| DRD2, block 2 (rs6265| rs11030101|rs1076563|rs1116313|rs2471857|rs2471854) | |||||
| CGCG | 87 / 328 | Ref | |||
| AACG | 62 / 183 | 1.21 | 1.25 | 0.94 – 1.65 | 0.12 |
| AATC | 47 / 105 | 1.48 | 1.50 | 1.10 – 2.03 | 0.01 |
| DRD2, block 3 (rs4245146|rs7122454|rs11214611) | |||||
| TGA | 79 / 287 | Ref | |||
| CGA | 86 / 215 | 1.33 | 1.29 | 1.00 – 1.66 | 0.05 |
| CCG | 31 / 113 | 1.00 | 1.01 | 0.69 – 1.48 | 0.95 |
| DRD3, block 1 (rs3732790|rs963468|rs324036) | |||||
| TGC | 91 / 311 | Ref | |||
| AAC | 73 / 219 | 1.10 | 1.08 | 0.85 – 1.38 | 0.52 |
| TGG | 30 / 75 | 1.26 | 1.26 | 0.90 – 1.74 | 0.17 |
| DRD3, block 3 (rs226082|rs324026) | |||||
| TT | 116 / 398 | Ref | |||
| CC | 72 / 194 | 1.20 | 1.23 | 0.97 – 1.56 | 0.09 |
| TC | 8 / 26 | 1.04 | 1.08 | 0.61 – 1.91 | 0.80 |
| BDNF (rs6265|rs11030101) | |||||
| CT | 77 / 202 | Ref | |||
| CA | 55 / 164 | 0.91 | 0.94 | 0.70 – 1.26 | 0.67 |
| TA | 40 / 76 | 1.25 | 1.27 | 0.93 – 1.72 | 0.13 |
Dysk / No Dysk: number of Parkinson's patients with dyskinesia / without dyskinesia. Ref: Reference. 95%CI: 95% confidence interval. NA: not available due to small number of subjects in cells. Adj. = adjusted analysis. Adjusted analysis was adjusted for PEG study, gender, disease duration, levodopa equivalent daily dosage, ethnicity, and age at diagnosis.
Table 3.
Poisson regression analyses for the association between ‘severity of dyskinesia’ and the haplotypes for the DRD1, DRD2, DRD3 and the BDNF loci in PD patients (N=418 patients) with a minimum of 3 years PD disease duration.
| N | Crude | Adj | |||
|---|---|---|---|---|---|
| No/Mild/ Severe | RR | RR | 95%CI | p-value | |
| DRD1 (rs4867798 rs5326|rs265981) | |||||
| TCA | 212/46/21 | Ref | |||
| TCG | 211 / 46/18 | 0.95 | 0.97 | 0.71 -1.33 | 0.85 |
| CCG | 106/19/15 | 1.11 | 1.15 | 0.78-1.70 | 0.48 |
| CTG | 87/21/9 | 1.06 | 1.14 | 0.77-1.69 | 0.52 |
| DRD2, block 1 (rs10891549|rs1554929|rs1124493|rs2242592|rs2245805) | |||||
| CTAGG | 293 / 55 / 23 | Ref | |||
| TCGTT | 183/44/16 | 1.15 | 1.17 | 0.87-1.57 | 0.29 |
| TCAGG | 103/27/21 | 1.68 | 1.62 | 1.14-2.30 | 0.01 |
| TCGTG | 30/2/3 | 0.84 | 0.74 | 0.28-2.01 | 0.56 |
| DRD2, block 2 (rs6265| rs11030101rs1076563|rs1116313|rs2471857|rs2471854) | |||||
| CGCG | 328 / 60 / 27 | Ref | |||
| AACG | 183/46/16 | 1.16 | 1.20 | 0.90-1.60 | 0.22 |
| AATC | 105/26/21 | 1.63 | 1.62 | 1.15-2.28 | 0.01 |
| DRD2, block 3 (rs4245146|rs7122454|rs11214611) | |||||
| TGA | 287 / 56 / 23 | Ref | |||
| CGA | 215/57/29 | 1.37 | 1.33 | 1.01 -1.74 | 0.04 |
| CCG | 113/19/12 | 1.07 | 1.08 | 0.72-1.63 | 0.71 |
| DRD3, block 1 (rs3732790|rs963468|rs324036) | |||||
| TGC | 311 / 64 / 27 | Ref | |||
| AAC | 219 / 49 / 24 | 1.13 | 1.11 | 0.86 – 1.43 | 0.42 |
| TGG | 75 / 18 / 12 | 1.36 | 1.35 | 0.95 – 1.92 | 0.10 |
| DRD3, block 3 (rs226082|rs324026) | |||||
| TT | 398 / 82 / 34 | Ref | |||
| CC | 194 / 46 / 26 | 1.26 | 1.29 | 1.00 – 1.65 | 0.05 |
| TC | 26 / 4 / 4 | 1.21 | 1.26 | 0.71 – 2.25 | 0.43 |
| BDNF (rs6265|rs11030101) | |||||
| CT | 202 / 54 / 23 | Ref | |||
| CA | 164 / 40 / 15 | 0.89 | 0.92 | 0.68 – 1.25 | 0.60 |
| TA | 76 / 22 / 18 | 1.40 | 1.43 | 1.01 – 2.03 | 0.04 |
No/Mild/Severe: Number of Patients with respectively no, mild or severe dyskinesia. N: Number. Ref: Reference. 95%CI: 95% confidence interval. NA: not available due to small number of subjects in cells. Adj. = adjusted analysis. Adjusted analysis was adjusted for PEG study, gender, disease duration, levodopa equivalent daily dosage, ethnicity, and age at diagnosis.
Combining “risk haplotypes” (defined as the haplotypes in each haploblock associated with increased risk for dyskinesia) into a risk-score for the DRD2 (see supplemental table e-2 for prevalence of risk-scores), we estimated a 16% risk increase for dyskinesia and 21% risk increase for ‘severity of dyskinesia’ with each unit increase in risk score. Analyzing the risk score as a categorical variable, there is also a clear increase in odds with an increasing risk score. Compared to no risk haplotypes, having 1 to 3 risk haplotypes increases the risk by about 55%, while having 4 or more haplotypes increases the risk by 153%. The risk score for the DRD3 showed an increase of 17% with each additional “risk haplotype” and a 20% increase for ‘severity of dyskinesia’, though both confidence intervals included or were very close to the null. However, we observed a dose-response pattern when reviewing the association with the risk score as a categorical variable (see table 4).
Table 4.
Poisson regression analyses with robust error variance and a log link function for the association between the DRD2 and DRD3 risk scores and dyskinesia.
| Number of Risk Haplotypes | Dyskinesia | Severity of dyskinesia | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Crude | Adj. | Crude | Adj. | |||||||
| RR | RR | 95%CI | p-value | RR | RR | 95%CI | p-value | |||
| DRD2, risk score | ||||||||||
| Continuous | 1.17 | 1.16 | 1.05 – 1.29 | 0.004 | 1.20 | 1.21 | 1.08 – 1.35 | 0.001 | ||
| N Dysk / No Dysk | N No/ Mild / Severe | |||||||||
| 0 | 20 / 103 | Ref | 103 / 15 / 5 | Ref | ||||||
| 1 to 3 | 61 / 181 | 1.55 | 1.54 | 0.98 – 2.42 | 0.06 | 181 / 41 / 20 | 1.76 | 1.61 | 1.00 – 2.60 | 0.05 |
| 4 to 6 | 17 / 25 | 2.49 | 2.53 | 1.48 – 4.34 | 0.001 | 25 / 10 / 7 | 3.62 | 2.80 | 1.58 – 4.93 | <0.001 |
| DRD3, risk score | ||||||||||
| Continuous | 1.16 | 1.17 | 0.99 – 1.40 | 0.07 | 1.22 | 1.20 | 1.00 – 1.45 | 0.04 | ||
| N Dysk / No Dysk | N No/ Mild / Severe | |||||||||
| 0 | 30 / 130 | Ref | 130 / 22 / 8 | Ref | ||||||
| 1 | 62 / 164 | 1.46 | 1.45 | 0.99 – 2.12 | 0.06 | 164 / 40 / 22 | 1.57 | 1.53 | 1.02 – 2.30 | 0.04 |
| 2 to 4 | 6 / 15 | 1.52 | 1.72 | 0.78 – 3.81 | 0.18 | 15 / 4 / 2 | 1.60 | 1.84 | 0.79 – 4.31 | 0.16 |
N Dysk / No Dysk: number of Parkinson's patients with dyskinesia / without dyskinesia. No/Mild/Severe: Number of Patients with respectively no, mild or severe dyskinesia. Ref: Reference category. 95%CI: 95% confidence interval. Adj. = adjusted analysis. Adjusted analysis was adjusted for PEG study, gender, disease duration, levodopa equivalent daily dosage, ethnicity, and age at diagnosis.
Restricting the sample to patients of European ancestry did not change the results for dyskinesia nor for ‘severity of dyskinesia’. None of the individual SNPs were responsible for the haplotype findings in the DRD loci. However, in BDNF, the rs6265 SNP was associated with dyskinesia (aRR: 1.31 (95%CI: 1.01-1.70; p:0.04)), while the rs11030101 was not (aRR: 0.96 (95%CI: 0.74-1.26); p:0.78). Risk among minor allele carriers of rs6265 was even more pronounced for ‘severity of dyskinesia’ (aRR: 1.43 (95%CI: 1.10-1.97); p:0.01).
Discussion
In our large community-based study of PD conducted in central California, we estimated an increased odds for developing dyskinesia after dopaminergic treatment in specific DRD2, DRD3, and BDNF but not DRD1 haplotype carriers. Our study finds that genetic variation in DRD2 influences the prevalence of dyskinesia and its severity in a dose response manner. Some variations in DRD3 and BDNF may further contribute to the risk and “severity of dyskinesia”. Our findings proved robust in sensitivity analyses. The three DRD2 “risk haplotypes” are frequently prevalent in the population (supplemental figure e-2 and table e-2). Hence, a large percentage of L-dopa treated patients are at increased risk of developing dyskinesia. Overall, about 30% of subjects have no DRD2 risk haplotype, with about 60% having one to three risk haplotypes; and about 10% carrying four or more risk haplotypes. Especially patients with more than three risk haplotypes are at a much higher risk since the risk of dyskinesia in medicated patients increased by 153%.
We chose to not use a correction for multiple testing, because the four genes in this study were chosen based on previous findings and because the haplotypes within each gene are highly correlated. However, when using an excessively restrictive Bonferroni correction assuming 10 tests (8 haplotypes and 2 risk scores) leading to an alpha value of 0.005, our findings for the DRD2 risk score would still be considered statistically significant. Only one previous study investigated an association between haplotypes in DRD2 and dyskinesia. The haplotypes in this relatively small Brazilian clinic-based population (N=199) were generated from different SNPs.[14] Our SNPs were chosen based on tagging characteristics rather than to replicate this previous study. Even though we cannot compare our results directly to these haplotypes, several SNPs are in LD with ours. Specifically, rs6277 and rs2734849 are associated with the rs1089154 and rs1554929 from our first haploblock (R-square is 1.0 for rs6277 and 0.77 for rs2734849) based on HapMap data. Furthermore, rs2283265, rs1076560 and rs1800497 are associated with the rs2471857 and rs2471854 from our second haploblock with R-squares of 0.98, 0.94 and 0.64, respectively. According to functional studies, the rs1800497 is associated with a decrease in DRD2 expression.[29] Thus, our haploblock may be associated with functional changes in DRD2 expression.
Previous studies reviewed the association between DRD3 and dyskinesia, focusing mainly on the rs6280 SNP (Ser9Gly).[11,13,18] The particular Ser9Gly polymorphism in the dopamine 3 receptor has a higher binding affinity to dopamine,[30] and is in high LD with our third haploblock (R-square for rs226082 and rs324026 are 0.84 and 1.00, respectively). Although one Korean study found an association with a small subgroup of patients with diphasic dyskinesia,[13] other studies showed no association with dyskinesia.[11,18] Unfortunately, we did not gather information about the type of dyskinesia and cannot stratify our patients according to subtypes. The estimated effect size for DRD3 in our study is smaller than for DRD2 limiting our power to identify a statistically significant finding. However, our findings suggest that the DRD3 haplotypes may be relevant for dyskinesia, even though they are not formally statistically significant.
In our study, the rs6265 SNP of BDNF was associated with dyskinesia while haplotype associations were not evident. The minor allele of SNP rs6265 (Val66Met) has been previously associated with time to development of dyskinesia in a study of 315 patients from the Cambridge Centre for Brain repair.[20] Yet another study involving 285 Australian patients with PD did not find any difference in time to onset of dyskinesia. [19] The minor allele of rs6265 is associated with a change of valine to methionine at codon 66 causing decreased protein secretion. This change has been used as a biologic argument for increased susceptibility for dyskinesia.[6]
In genetic studies, an important potential confounder is race/ethnicity. To avoid potential population stratification, we limited our analyses to patients from European ancestry in sensitivity analyses and our findings did not change. We adjusted our analysis for race/ethnicity according to three categories. Primary analyses were not adjusted for population substructure using fractional ancestry because AIMs were not available on ∼25% of the individuals. However, when we limited our analysis to those individuals with AIMs and adjusted for population structure, we got essentially the same results (results not shown). Race/ethnicity in our study was based mostly on genetic AIMS markers, or – if unavailable – on self-report of parental origin. The correlation between these two measures is high (95.4%) in our population. Although there may be some measurement error for race/ethnicity, based on self-report, there is no indication this would unduly influence our findings.
Information on dyskinesia on almost all subjects was available though we did not assess dyskinesia during the baseline examination early in disease for subjects enrolled in PEG1. A third of all patients were lost to follow up between baseline and follow-up exam in PEG1. The main reason for this was mortality and disease severity preventing participation. Thus, if both dyskinesia and genetic variants of interest influence mortality, we may have a survival bias in PEG1 subjects. However, as findings were consistent across both studies (PEG1 and PEG2) and dyskinesia was assessed in PEG2 subjects at baseline such a survival bias appears unlikely. We only gathered information about dyskinesia prevalence at the time of motor function assessment during physical examinations and we did not record the exact start time of dyskinesia leading to left censoring. This limitation does not allow us to conduct a time-to-event analysis. However, most of the PEG patients were early in their presentation (on average 5.1 years at first dyskinesia assessment) and were followed up within 2-5 years of the first dyskinesia assessment.
The presence of one of the risk haplotypes, or variant alleles, can have major implications for best practices in treatment of PD. Patients who are at higher risk for developing dyskinesia may benefit from therapy with dopamine agonists along with a more strict L-dopa-sparing strategy. Treatment plans should reflect the difficulty in managing PD, to maximize optimal effects (L-dopa ‘on’ time) but also mitigate dyskinesia risk. Our results suggest that individual treatment plans should consider a patient's genetic profile.
Conclusion
Several haplotypes in DRD2, possibly haplotypes in DRD3 and the minor allele of rs6265 in BDNF, increased the risk of dyskinesia in our study. Levodopa induced dyskinesia and PD symptoms must be approached as a tradeoff. Nevertheless, genetic information may help prevent or postpone this debilitating consequence of treatment and may improve patient-centered, personalized therapy. Association studies require confirmation and the health care economy of implementing more personalized treatment should also be quantified before decisions are made. PD patients with these specific risk haplotypes may be prime candidates for testing approaches to prevent or delay the development of dyskinesia.
Supplementary Material
Supplemental figure e-1. Patient recruitment, in- and exclusion-criteria, flow diagram of the two PEG studies (N=418)
Supplemental figure e-2. Linkage Disequilibrium (LD) plots for the DRD-loci in our study population, LD measured with R-squared. Each LD plot shows the various haplotypes in a haploblock with its population frequency and connections. In the crossing areas, the value of multiallelic D' is shown. This represents the level of recombination between the two blocks.[25]
Highlights.
- Dopamine receptor genes (DRD1-3) and Brain Derived Neurotrophic Factor (BDNF) have been proposed to be involved in the development of dyskinesia
- DRD2 haplotypes, and potentially DRD3 haplotypes and rs6265 BDNF-SNP, were associated with dyskinesia
- The subpopulation of patients with these genetic markers appear prime candidate for testing approaches to prevent or delay the development of dyskinesia
Acknowledgments
We would like to acknowledge the work performed by Stephanie F Bortnick (UBC) for the genotyping; Tara Candido for overseeing and managing the combined effort of UBC with UCLA; and Prof. Yvette Bordelon for her work as a movement specialist assessing the patients in this study. In addition, we would like to thank all the participants in this study.
Funding sources: This project has been made available with funding from National Institutes of Health (grants: R01-ES010544, U54-ES012078, P01-ES016732, and P50-NS038367; initial pilot funding (P30-ES07048); NIH Training Grant in Genomic Analysis and Interpretation (T32HG002536)); Burroughs Wellcome Fund fellowships; a Community Fast Track grant by the MJFox Foundation; a pilot grant by the American Parkinson Disease Association; and Canada Excellence Research Chairs program. Leading Edge Endowment Funds provided by the Province of British Columbia, LifeLabs, and Genome BC support the Dr. Donald Rix BC Leadership Chair, Canadian Federal Government, and Cunhill Foundation.
In addition, M. Farrer has a patent on genetic variability in LRRK2 and Parkinson's disease (US8409809, US8455243B2) and on LRRK2 mouse models that have been subsequently developed, with royalties paid through Mayo Foundation.
Footnotes
Author Contributions: Jeff M Bronstein has evaluated the majority of patients to evaluate their Parkinson's disease status and symptoms; Ilaria Guella and Matt J Farrer were responsible for the genetic analysis and quality control; Matt J Farrer and Beate R Ritz were responsible for the conception, and overseeing the organization and execution; and the analysis were primarily performed by Cynthia DJ Kusters with the help of Kimberly C Paul, Janet S Sinsheimer and Beate R Ritz. The first draft of the article was written by Cynthia DJ Kusters. All of the authors contributed to the inference of the results from the analysis, the editing and writing of the paper.
Conflicts of Interest: None
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
References
- 1.Factors impacting on quality of life in Parkinson's disease: results from an international survey. Mov Disord. 2002;17:60–7. doi: 10.1002/mds.10010. [DOI] [PubMed] [Google Scholar]
- 2.Bjornestad A, Forsaa EB, Pedersen KF, Tysnes OB, Larsen JP, Alves G. Risk and course of motor complications in a population-based incident Parkinson's disease cohort. Parkinsonism Relat Disord. 2016;22:48–53. doi: 10.1016/j.parkreldis.2015.11.007. [DOI] [PubMed] [Google Scholar]
- 3.Fabbrini G, Brotchie JM, Grandas F, Nomoto M, Goetz CG. Levodopa-induced dyskinesias. Mov Disord. 2007;22:1379–89. doi: 10.1002/mds.21475. quiz 1523. [DOI] [PubMed] [Google Scholar]
- 4.Péchevis M, Clarke CE, Vieregge P, Khoshnood B, Deschaseaux-Voinet C, Berdeaux G, Ziegler M. Effects of dyskinesias in Parkinson's disease on quality of life and health-related costs: a prospective European study. Eur J Neurol. 2005;12:956–63. doi: 10.1111/j.1468-1331.2005.01096.x. [DOI] [PubMed] [Google Scholar]
- 5.Kalinderi K, Fidani L, Katsarou Z, Bostantjopoulou S. Pharmacological treatment and the prospect of pharmacogenetics in Parkinson's disease. Int J Clin Pract. 2011;65:1289–94. doi: 10.1111/j.1742-1241.2011.02793.x. [DOI] [PubMed] [Google Scholar]
- 6.Droździk M, Białecka M, Kurzawski M. Pharmacogenetics of Parkinson's disease - through mechanisms of drug actions. Curr Genomics. 2013;14:568–77. doi: 10.2174/1389202914666131210212521. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Fumagalli F, Racagni G, Riva MA. Shedding light into the role of BDNF in the pharmacotherapy of Parkinson's disease. Pharmacogenomics J. 6:95–104. doi: 10.1038/sj.tpj.6500360. [DOI] [PubMed] [Google Scholar]
- 8.Guillin O, Diaz J, Carroll P, Griffon N, Schwartz JC, Sokoloff P. BDNF controls dopamine D3 receptor expression and triggers behavioural sensitization. Nature. 2001;411:86–9. doi: 10.1038/35075076. [DOI] [PubMed] [Google Scholar]
- 9.Kaplan N, Vituri A, Korczyn AD, Cohen OS, Inzelberg R, Yahalom G, Kozlova E, Milgrom R, Laitman Y, Friedman E, Rosset S, Hassin-Baer S. Sequence variants in SLC6A3, DRD2, and BDNF genes and time to levodopa-induced dyskinesias in Parkinson's disease. J Mol Neurosci. 2014;53:183–188. doi: 10.1007/s12031-014-0276-9. [DOI] [PubMed] [Google Scholar]
- 10.Oliveri RL, Annesi G, Zappia M, Civitelli D, Montesanti R, Branca D, Nicoletti G, Spadafora P, Pasqua aa, Cittadella R, Andreoli V, Gambardella a, Aguglia U, Quattrone a. Dopamine D2 receptor gene polymorphism and the risk of levodopa-induced dyskinesias in PD. Neurology. 1999;53:1425–30. doi: 10.1212/WNL.53.7.1425. [DOI] [PubMed] [Google Scholar]
- 11.Kaiser R, Hofer A, Grapengiesser A, Gasser T, Kupsch A, Roots I, Brockmoller J. L-Dopa-induced adverse effects in PD and dopamine transporter gene polymorphism. Neurology. 2003;60:1750–1755. doi: 10.1212/01.wnl.0000068009.32067.a1. [DOI] [PubMed] [Google Scholar]
- 12.Greenbaum L, Goldwurm S, Zozulinsky P, Lifschytz T, Cohen OS, Yahalom G, Cilia R, Tesei S, Asselta R, Inzelberg R, Kohn Y, Hassin-Baer S, Lerer B. Do tardive dyskinesia and L-dopa induced dyskinesia share common genetic risk factors? An exploratory study. J Mol Neurosci. 2013;51:380–8. doi: 10.1007/s12031-013-0020-x. [DOI] [PubMed] [Google Scholar]
- 13.Lee JY, Cho J, Lee EK, Park SS, Jeon BS. Differential genetic susceptibility in diphasic and peak-dose dyskinesias in Parkinson's disease. Mov Disord. 2011;26:73–9. doi: 10.1002/mds.23400. [DOI] [PubMed] [Google Scholar]
- 14.Rieck M, Schumacher-Schuh AF, Altmann V, Francisconi CL, Fagundes PT, Monte TL, Callegari-Jacques SM, Rieder CR, Hutz MH. DRD2 haplotype is associated with dyskinesia induced by levodopa therapy in Parkinson's disease patients. Pharmacogenomics. 2012;13:1701–10. doi: 10.2217/pgs.12.149. [DOI] [PubMed] [Google Scholar]
- 15.Strong JA, Dalvi A, Revilla FJ, Sahay A, Samaha FJ, Welge JA, Gong J, Gartner M, Yue X, Yu L. Genotype and smoking history affect risk of levodopa-induced dyskinesias in Parkinson's disease. Mov Disord. 2006;21:654–9. doi: 10.1002/mds.20785. [DOI] [PubMed] [Google Scholar]
- 16.Zappia M, Annesi G, Nicoletti G, Arabia G, Annesi F, Messina D, Pugliese P, Spadafora P, Tarantino P, Carrideo S, Civitelli D, De Marco EV, Cirò-Candiano IC, Gambardella A, Quattrone A. Sex differences in clinical and genetic determinants of levodopa peak-dose dyskinesias in Parkinson disease: an exploratory study. Arch Neurol. 2005;62:601–5. doi: 10.1001/archneur.62.4.601. [DOI] [PubMed] [Google Scholar]
- 17.Wang J, Liu ZL, Chen B. Association study of dopamine D2, D3 receptor gene polymorphisms with motor fluctuations in PD. Neurology. 2001;56:1757–9. doi: 10.1212/wnl.56.12.1757. [DOI] [PubMed] [Google Scholar]
- 18.Paus S, Gadow F, Knapp M, Klein C, Klockgether T, Wüllner U. Motor complications in patients form the German Competence Network on Parkinson's disease and the DRD3 Ser9Gly polymorphism. Mov Disord. 2009;24:1080–4. doi: 10.1002/mds.22508. [DOI] [PubMed] [Google Scholar]
- 19.Cheshire P, Bertram K, Ling H, O'Sullivan SS, Halliday G, McLean C, Bras J, Foltynie T, Storey E, Williams DR. Influence of single nucleotide polymorphisms in COMT, MAO-A and BDNF genes on dyskinesias and levodopa use in Parkinson's disease. Neurodegener Dis. 2013;13:24–28. doi: 10.1159/000351097. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Foltynie T, Cheeran B, Williams-Gray CH, Edwards MJ, Schneider SA, Weinberger D, Rothwell JC, Barker RA, Bhatia KP. BDNF val66met influences time to onset of levodopa induced dyskinesia in Parkinson's disease. J Neurol Neurosurg Psychiatry. 2009;80:141–4. doi: 10.1136/jnnp.2008.154294. [DOI] [PubMed] [Google Scholar]
- 21.Hughes AJ, Ben-Shlomo Y, Daniel SE, Lees AJ. What features improve the accuracy of clinical diagnosis in Parkinson's disease: a clinicopathologic study. Neurology. 1992;42:1142–6. doi: 10.1212/wnl.42.6.1142. [DOI] [PubMed] [Google Scholar]
- 22.Ritz B, Rhodes SL, Bordelon Y, Bronstein J. α-Synuclein genetic variants predict faster motor symptom progression in idiopathic Parkinson disease. PLoS One. 2012;7:e36199. doi: 10.1371/journal.pone.0036199. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Goetz CG, Tilley BC, Shaftman SR, Stebbins GT, Fahn S, Martinez-Martin P, Poewe W, Sampaio C, Stern MB, Dodel R, Dubois B, Holloway R, Jankovic J, Kulisevsky J, Lang AE, Lees A, Leurgans S, LeWitt PA, Nyenhuis D, Olanow CW, Rascol O, Schrag A, Teresi JA, van Hilten JJ, LaPelle N. Movement Disorder Society UPDRS Revision Task Force, Movement Disorder Society-sponsored revision of the Unified Parkinson's Disease Rating Scale (MDS-UPDRS): scale presentation and clinimetric testing results. Mov Disord. 2008;23:2129–70. doi: 10.1002/mds.22340. [DOI] [PubMed] [Google Scholar]
- 24.Tomlinson CL, Stowe R, Patel S, Rick C, Gray R, Clarke CE. Systematic review of levodopa dose equivalency reporting in Parkinson's disease. Mov Disord. 2010;25:2649–53. doi: 10.1002/mds.23429. [DOI] [PubMed] [Google Scholar]
- 25.Barrett JC, Fry B, Maller J, Daly MJ. Haploview: analysis and visualization of LD and haplotype maps. Bioinformatics. 2005;21:263–5. doi: 10.1093/bioinformatics/bth457. [DOI] [PubMed] [Google Scholar]
- 26.Gabriel SB. The Structure of Haplotype Blocks in the Human Genome. Science (80-) 2002;296:2225–2229. doi: 10.1126/science.1069424. [DOI] [PubMed] [Google Scholar]
- 27.Stephens M, Smith NJ, Donnelly P. A New Statistical Method for Haplotype Reconstruction from Population Data. Am J Hum Genet. 2001;68:978–989. doi: 10.1086/319501. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Fang J. Using SAS ® Procedures FREQ, GENMOD, LOGISTIC, and PHREG to Estimate Adjusted Relative Risks – A Case Study. SAS Glob Forum. 2011;2011:345. [Google Scholar]
- 29.Eisenstein SA, Bogdan R, Love-Gregory L, Corral-Frías NS, Koller JM, Black KJ, Moerlein SM, Perlmutter JS, Barch DM, Hershey T. Prediction of striatal D2 receptor binding by DRD2/ANKK1 TaqIA allele status. Synapse. 2016;70:418–31. doi: 10.1002/syn.21916. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Lundstrom K, Turpin MP. Proposed Schizophrenia-Related Gene Polymorphism: Expression of the Ser9Gly Mutant Human Dopamine D 3 Receptor with the Semliki Forest Virus System. Biochem Biophys Res Commun. 1996;225:1068–1072. doi: 10.1006/bbrc.1996.1296. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Supplemental figure e-1. Patient recruitment, in- and exclusion-criteria, flow diagram of the two PEG studies (N=418)
Supplemental figure e-2. Linkage Disequilibrium (LD) plots for the DRD-loci in our study population, LD measured with R-squared. Each LD plot shows the various haplotypes in a haploblock with its population frequency and connections. In the crossing areas, the value of multiallelic D' is shown. This represents the level of recombination between the two blocks.[25]