Effect of GBA Mutations on Phenotype of Parkinson's Disease: A Study on Chinese Population and a Meta-Analysis

GBA has been identified as a genetic risk factor for PD. Whether the clinical manifestations of PD patients with or without GBA mutations are different has still not reached a consensus. We firstly detected the GBA mutation L444P in 1147 Chinese PD patients and simultaneously evaluated their corresponding clinical data. Then we compared the phenotypes between 646 PD patients with GBA mutations and 10344 PD patients without GBA mutations worldwide through meta-analysis. Through the method of meta-analysis, there was significant difference in age at onset (MD = −3.10 [95% CI: −4.88, −1.32]), bradykinesia as an initial symptom (OR = 1.49 [95% CI: 1.15, 1.94]), having family history (OR = 1.50 [95% CI: 1.18, 1.91]), and dementia (OR = 3.21 [95% CI: 1.97, 5.24]) during the comparison between PD patients with and without GBA mutations. While, in the aspect of tremor as an initial symptom (OR = 0.81 [95% CI: 0.64, 1.03]), the severity of motor symptoms such as H-Y (MD = 0.06 [95% CI: −0.06, 0.17]) and UPDRS-III (MD = 1.61 [95% CI: −0.65, 3.87]) and having dyskinesia (OR = 1.60 [95% CI: 0.90, 2.84]) during the comparison between the two groups revealed no statistical differences. Our results suggested that the phenotypes of PD patients with GBA mutations are different from GBA noncarriers.


Introduction
Parkinson's disease (PD) is the second most common progressive neurodegenerative disorder. Though the etiology of PD remains unclear, there is an increasing evidence that genetic-factor contributes to the etiology of PD.
Mutations in the gene encoding the lysosomal enzyme glucocerebrosidase (GBA) have been identified as a genetic risk factor for PD [1]. Aharon-Peretz and her colleagues [2] reported that the overall clinical manifestations and age at disease onset have no differences in PD patients with GBA mutations (GBA + PD) compared with those without mutations (GBA − PD) in 148 Ashkenazi PD patients. However, recent researches have shown that the clinical features of GBA + PD differ from GBA − PD to some extent. Winder-Rhodes et al. [3] showed that GBA carriers in PD were inclined to suffer an earlier age at onset (AAO) and more severe nonmotor and motor symptoms. Hu and other researchers [4] reported that GBA mutations influenced the course of PD with respect to the appearance of dementia. Whether the clinical manifestations of PD patients with or without GBA mutations are different or not has still not reached a consensus.
In order to evaluate the effect of GBA on the phenotype of PD, we firstly explored the relationship between GBA mutations and their clinical characteristics in Chinese PD patients. Then the method of meta-analysis was used to assess the possible role of GBA in the phenotype of PD with larger sample size worldwide. (i) case reports, reviews, or editorials; (ii) functional researches; (iii) overlapping study group about the same study outcomes; (iv) incomplete data.

Genetic Analysis.
All 1147 PD patients were detected for GBA gene L444P mutation, the most common PD-associated GBA mutation in Chinese population [24]. The screening procedures and details have been reported recently [25].

Meta-Analysis.
Including our own data of Chinese population, a meta-analysis related to the above topic was conducted. Eligible studies had to meet the following criteria: (1) being a case-control study except for reviews, case reports, editorials, or functional researches; (2) all PD patients being diagnosed according to UKBB criteria with the exception that a positive family history was not a part of the exclusion criteria; (3) only including publications related to GBA mutation analysis; (4) clearly reporting results of GBA mutations and corresponding clinical data. Then we searched electronic databases including Embase, PubMed, Cochrane Library, and Web of Knowledge and Wanfang database and CNKI up to May 1, 2015, using combination of following keywords:    GBA, glucocerebrosidase, and Parkinso * both in English and in Chinese. Reference lists and personal communications of authors were also referred to as sources to include articles cited elsewhere.
To select studies for further assessment, two authors independently scanned the abstracts, titles, or both sections of each retrieved record. All potentially relevant articles were investigated in full text. For studies satisfying the aforesaid criteria, two authors independently abstracted the following data: year of publication, first author's surname, country of participants, numbers of PD patients with and without GBA mutations, and corresponding clinical information.  The flowchart of studies selection and reasons for exclusion are presented in Figure 1. The qualities of the included studies were evaluated by the Newcastle-Ottawa Scale (NOS) [26]. In order to assess the strength of association between GBA mutations and clinical manifestation, dichotomous outcome was expressed as odds ratio (OR) while continuous outcome was expressed as mean difference (MD) with 95% confidence intervals (CI). Heterogeneity across individual studies was identified by a standard test with a significance level of = 0.1 and 2 . If heterogeneity did not exist ( > 0.10) or the severity of heterogeneity was accepted ( 2 ≤ 50%), the fixed effect model was adopted to calculate the pooled OR and MD value. Otherwise, the random effect model was used. Funnel-plot analysis was used to assess the reporting bias. All analyses were carried out by using the Review Manager software package v. 5

Results
34 GBA + PD and 1113 GBA − PD were available in our own data from Chinese. A total of 18 eligible studies were included during the process of meta-analysis. The main characteristics of all included studies are summarized in Table 1.

Onset Characteristics
3.1.1. Family History. PD patients with family history are defined as having at least one first-or second-degree relative with the diagnosis of PD. There are 38 PD patients (1 GBA + PD, 37 GBA − PD) with family history in our own data. We included another 9 publications to assess the relationship between GBA status and family history of PD patients. The pooled OR of family history in GBA + PD and GBA − PD was 1.50 [95% CI: 1.18, 1.91] (Figure 2(a)). That means GBA + PD was more likely to be exposed to family history than GBA − PD, even though the relationship was negative in our own data. The funnel plot was symmetric (Figure 2(b)).  (Figure 3(a)). It means that the AAO is nearly 3 years earlier in GBA + PD than GBA − PD. The funnel plot was asymmetric and the bias leaned to no difference in the two groups (Figure 3(b)). The AAO of GBA + PD and GBA − PD are 50.18 ± 9.44 and 54.74 ± 11.52 individually in our own data from Chinese population, which was consistent with the results of meta-analysis.
Unfortunately, there was no significant difference in our own data from Chinese population, which was consistent with publications that referred to Chinese population [6,20].   [14,17] and MMSE [8,9] or reported by author [27] in included publication. Even though all the dementia statuses were defined by various rating scales, the heterogeneity of all included publications was accepted when we made the metaanalysis. The pooled OR of dementia between GBA + PD and GBA − PD was 3.21 [95% CI: 1.97, 5.24] (Figure 6(a)) and the funnel plot was symmetric (Figure 6(b)). The result means that PD patients with GBA mutations are 3 times more likely than uncarriers to develop dementia.  (Figure 7(a)). The funnel plot was symmetric (Figure 7(b)).

Discussion
We firstly presented an extensive and detailed phenotype description of GBA + PD in Chinese together with metaanalysis worldwide, focusing on not only the disease onset features but also disease progression characteristics. The phenotype of GBA + PD, sharing a spectrum of Parkinsonian phenotype, differs from GBA − PD in the following aspects: AAO, initial motor symptom, presentation with family history, and dementia. Totally, 646 GBA + PD and 10344 GBA − PD were included. The NOS scores of all included publications were rated from 6 stars to 9 stars, which showed that none of the included publications were of low quality.
Additionally, there was no obvious publication reporting bias from the symmetric funnel plots related to the analysis in family history, bradykinesia or tremor as an initial symptom, dementia, and dyskinesia. Even though there was no different in the funnel plot of AAO, we still figured out the earlier onset age of GBA + PD than that of GBA − PD. Thus, we can conclude that the GBA + PD patients are more likely to onset with earlier age than GBA − PD, which are easily ignored by clinical doctors because PD is more likely to develop in the old. In the same way, the MDs of H-Y and UPDRS -III showed no significant differences with so many positive reports. That is to say, GBA mutations in PD patients are not related to the severity in PD.
What is more, both genetic and clinical heterogeneities exist during the analysis process. All heterogeneities may originate from the following parts: firstly, the frequency of the different GBA mutations varies according to ethnicity [27]. For instance, in Ashkenazi Jewish (AJ), N370S is the most frequent mutation, whereas the L444P is more common in Asian population than others. Still, other variations, like R120W and D409H, were also reported in several studies but rarely showed positive outcomes. We could not extract single GBA mutation and corresponding clinical information from included publications. Thus, mixed GBA carriers and their clinical data together may affect the final results, especially for the ethnicity related data. Even in the same ethnicity, the heterogeneity also could not be avoided due to different genetic screening methods, other susceptible genes apart from GBA, and different degree of effects by the different GBA mutations [28]. Secondly, studies which did not adjust for age, sex, disease duration, environmental exposures, and other cofactors may influence clinical phenotype of PD.
The foundation underlying the relation of GBA genotype to clinical characteristics of PD remains elusive. Mutant alleles of GBA can result in widespread deficiency of the enzymatic activity that might be involved in abnormal synuclein aggregation [1]. Besides, both histopathologic and positron emission tomography studies have suggested that GBA carriers were significantly more likely than noncarriers to have diffused Lewy bodies which might be associated with a higher prevalence and severity of cognitive impairment and neuropsychiatric characteristics [29,30].

Conclusion
In conclusion, our results suggest that PD phenotype of GBA mutation carriers is more likely to appear with family history, earlier AAO, bradykinesia as an initial symptom, and presentation with dementia compared with those uncarriers. Further studies, especially cofactors, matched single mutation of GBA, and its correspondent clinical data are needed to further illustrate the relationship.