Bacterial, viral, and fungal infection‐related risk of Parkinson's disease: Meta‐analysis of cohort and case–control studies

Abstract Aims Recent studies showed that patients with various bacterial, viral, and fungal infections might be at increased risk of Parkinson's disease (PD). However, the risk of PD in patients with each specific infection varied. This meta‐analysis estimated the association between various infections and PD risk. Methods Literature published from January 1965 to October 2019 in PubMed and EMBASE databases was searched. Data were extracted and pooled using random/fixed effects model. Sensitivity analysis and meta‐regression were also performed to analyze the source of heterogeneity. Publication bias was estimated by the trim and fill. Results Twenty‐three out of 6,609 studies were included. Helicobacter pylori (HP; pooled OR = 1.653, 1.426–1.915, p < .001), hepatitis C virus (HCV; pooled OR = 1.195, 1.012–1.410, p = .035), Malassezia (pooled OR = 1.694, 1.367–2.100, p < .001), and pneumoniae (pooled OR = 1.595, 1.020–2.493, p = .041) infection were associated with increased PD risk. Influenza virus, herpes virus, hepatitis B virus, scarlet fever, mumps virus, chicken pox, pertussis, German measles, and measles were not associated with PD risk. After antiviral treatment against HCV reduced the risk of PD in patients with HCV infection (OR = 0.672, 0.571–0.791, p < .001). Significant heterogeneity exists among the included studies. Conclusion Patients with infection of HP, HCV, Malassezia, pneumoniae might be an increased risk of PD. Antiviral treatment of HCV could reduce the risk of PD.


| INTRODUC TI ON
Parkinson's disease (PD) is a common neurodegenerative disease in the elderly, which is clinically characterized by resting tremor, bradykinesia, rigidity, and postural instability and difficulty in walking (Gerfen, 2000). Approximately 7.5 million people worldwide are affected by Parkinson's disease and the prevalence increases rapidly with age. It has been anticipated that there will be nine million PD patients by 2030 (Pringsheim, Jette, Frolkis, & Steeves 2014;Ross & Abbott, 2014). The main pathological change of PD is degeneration and death of dopaminergic neurons in the substantia nigra due to unclear etiology and pathogenesis (Li et al., 2019;Salamon, Zádori, Szpisjak, Klivényi, & Vécsei, 2019). Similar to other neurodegenerative diseases, multiple factors, including gene, neuroinflammation, trauma, drugs, and toxicity, appear to play important roles in the development of PD (Dick et al., 2007;Mcgeer, Yasojima, & Mcgeer, 2003;Park et al., 2019;Xu, Chen, Xu, Zhang, & Li, 2018;Xu et al., 2014). Recently, infection is increasingly recognized as a risk factor for PD (Liu, Gao, & Hong, 2003;Mattson, 2004) because it may trigger chronic inflammation of the microglia (Alam et al., 2016) and, thus, may promote the onset of PD.
Notably, although there is no effective prevention strategy for PD, antiviral therapy against HCV has been reported to reduce the risk of PD in patients with HCV infection Su et al., 2019), indicating that treatment against pathogenic microorganisms may be a potentially effective method to prevent PD. However, whether treatment against pathogenic microorganisms other than HCV reduces the risk of PD has not been researched.
This meta-analysis analyzed the relationship between infection of 13 pathogenic microorganisms and risk of PD, including herpes virus, HBV, pertussis, scarlet fever, influenza, mumps, chicken pox, measles, Malassezia, HCV, HP, pneumonia and German measles. We also analyzed the preventive effect of antiviral therapy against HCV on risk of PD in patients with HCV infection. ). EMBASE and CNKI database was also searched for articles published using similar searching strategy. If any pathogenic microorganism was identified associated with PD in articles searched using above strategy, a manual search for the related literature between PD and this pathogenic microorganism species was performed. For example, we identified HP and HCV infection were reported associated with PD using above terms; then, we searched the databases manually using the following terms: ((Parkinson's disease) AND Helicobacter pylori), ((Parkinson's disease) AND Hepatitis C virus). Finally, the reference lists of the included articles were also searched manually to identify additional relevant studies not captured by our database search. We did not contact the authors for unpublished data.

| Inclusion and exclusion criteria
The inclusion and exclusion criteria were as follows: (a) Original studies determining relationship between infection and risk of PD published in English or Chinese, including case-control studies, and cohort studies were included for further evaluation, experiment, or studies on animals were excluded; (b) odds radios (ORs; unadjusted), relative risks (RRs), and hazard ratios (HRs) with 95% confidence intervals (CI) should be provided or could be calculated in included studies.

| Quality assessment of included studies
The quality of each included study was assessed separately by two investigators (H.W. and L.M.) using the validated Newcastle-Ottawa quality assessment scale (NOS; Stang, 2010) in the following three aspects: (a) selection of the participants, (b) comparability between the groups, and (c) ascertainment of the exposure of interest for case-control studies or the outcome of interest for cohort studies.
Disagreements of assessment between the two investigators were resolved by the third investigator (X.L.).

| Data extraction
A standardized data collection form was used to extracted the following data from each study: (a) last name of the first author, (b) publication year, (c) number of people in PD group and control group,

| Statistical analysis
All statistical analyses were performed using Stata 12.0 software (StataCorp LP). p < .05 was defined as statistically significant. Pooled OR with 95% CI was calculated using random/fixed effect model to assess the association of bacteria, virus, and fungus infections with the risks of PD. If I 2 < 50%, the fixed effects model was selected; otherwise, the random effects model was selected. Heterogeneity was determined by the I 2 statistic (I 2 > 50% was considered with significant heterogeneity). The contribution of publication year, mean age and/or age range, study design, PD diagnostic criteria, infected pathogenic microorganisms, methods for detection of infection, and samples used to detect infection to heterogeneity were analyzed by sensitivity analysis. If sensitivity analysis failed to identify the source of heterogeneity, and if there are more than three articles on this pathogenic microorganism, meta-regression was performed to identify the source of heterogeneity. Trim and fill method was used to assess the publication bias.

| Literature quality assessment using NOS
Each included study was assessed using NOS; all studies have a NOS score more than 7, indicating a favorable quality of included studies.
Detailed NOS score of each included study was listed in Table 2.

| Infection of influenza virus, herpes virus, HBV, scarlet fever, mumps, chicken pox, pertussis, German measles, and measles virus were not associated with risk of PD
The meta-analysis found that influenza virus (pooled OR = 1.

| Sensitivity analysis and meta-regression
No significant heterogeneity exists among studies on HP, Malassezia, pneumoniae, chicken pox, and German measles.
Due to the high heterogeneity when pooled ORs of HBV, HCV, influenza virus, mumps, herpes virus, and measles, we performed sensitivity analysis to identify the source of heterogeneity.
Sensitivity analysis found that excluding the study by Vlajinac et al. (2013) from the pooled analysis could significant reduce the heterogeneity of studies on influenza virus, mumps, and measles. This may be because that Vlajinac's study (Vlajinac et al., 2013) used a structured questionnaire to gained personal and family histories, which may affect the accuracy of infection history. After excluding Vlajinac's study (Vlajinac et al., 2013), the pooled OR of influenza virus, herpes virus, and mumps was not significantly changed (in- Significant heterogeneity was identified among studies on HCV (I 2 = 79%, p < .001), and sensitivity analysis found that Golabi's study (Golabi et al., 2017) accounted for the source of heterogeneity, excluding Golabi's study (Golabi et  For scarlet fever and pertussis, only two studies were included, which is insufficient for sensitivity analysis or meta-regression.

| Analysis of publication bias
The results of trim and fill method were shown in    (Vlajinac et al., 2013). Additionally, this meta-analysis, antiviral therapy against HCV could reduce the risk of PD. The NOS score of each included study was more than 7, indicating favorable quality of included studies.

| D ISCUSS I ON
In this meta-analysis, we found that infection of HP, HCV, Malassezia, and pneumoniae was positively associated with the risk of PD. HP infection has been found to increase the synthesis of MPTP or MPTP-like substance (Altschuler, 1996) and cause chronic inflammation in central nervous system which damage dopaminergic neurons (Hirai et al., 1995) via activating microglia (Streit, Mrak, & Griffin, 2004), releasing neurotoxic substances (Villarán et al., 2010), or inducing autoimmune responses . HP infection may also affect symptoms of PD via decreasing absorption of levodopa and was related to poorer motor function in PD patients (Shen et al., 2017;Suwarnalata et al., 2016). Therefore, HP infection may be a potential causal factor of PD onset. In clinic, it may be reasonable to consider screening and eradicating HP in patients with family history of PD or at high risk of PD, especially considering the high prevalence of HP infection.
In patients with PD, eradication of HP may alleviate motor symptoms or strengthen effect of levodopa, but whether eradicating HP affect the natural process or progression of PD remains to be further researched.
Hepatitis C virus has been reported to increase risk of PD (Kim et al., 2016); a previous meta-analysis also reported increased PD incidence in patients with HCV infection (Wijarnpreecha et al., 2018). HCV has been reported to cause PD by inducing inflammatory cytokine release and damaging dopaminergic neurons (Alam et al., 2016;Mattson, 2004). It has been reported that the essential HCV receptors such as CD81, claudin-1, occludin, LDLR, and scavenger receptor-B1 are expressed on brain microvascular endothelial cells, a major component of the blood-brain barrier, suggesting that HCV may infect the central nervous system through these receptors (Alam et al., 2016;Fletcher et al., 2012). HCV-induced inflammatory cytokines release may also contribute to the pathogenesis of PD. In animal models, HCV induced 60% of dopaminergic neuron death in rat midbrain (Alam et al., 2016). In patients, the toxic effect of HCV on dopaminergic neurons was found similar to 1-methyl-4-phenylpyridinium (MPP+), and increased the risk of PD (Alam et al., 2016). Our meta-analysis showed that the risk of PD in HCV patients received effective antiviral treatment against HCV is lower than those who did not, supporting that HCV may be a risk factor for PD Su et al., 2019) and that antiviral treatment against HCV could reduce the risk of PD . Therefore, effective and more active antiviral treatment should be considered in HCV patients; the association between Notably, reports showed that receiving interferon-based antiviral therapy for HCV increased the risk of PD, this may be due to increased drug-induced parkinsonism in patients receiving interferon therapy .
In our meta-analysis, Malassezia infection was related to in- Similar to Malassezia infection, our meta-analysis and other reports also related Pneumoniae infection to increased risk of PD (Bu et al., 2015). However, these results may be attributed to increased pneumonia incidence in PD patients due to oropharyngeal dysphagia-induced aspiration or movement disorder-induced hypostatic pneumonia (Mamolar Andrés et al., 2017;Miyazaki, Arakawa, & Kizu, 2002). Whether infection of pneumoniae could increase the risk of PD still needs to be further investigated.
Additionally, previous studies reported that infection of pertussis, scarlet fever, HBV, herpes virus, influenza virus, mumps, and measles increased the risk of PD (Bu et al., 2015;Harris et al., 2012;Hemling et al., 2003;Sasco & Paffenbarger, 1985;Vlajinac et al., 2013); however, our meta-analysis did not found significant association between infection of these pathogenic microorganisms and risk of PD possibly due to the limited number of studies. The relation between these pathogenic microorganisms and risk of PD needs to be further studied.
Significant heterogeneity exists in meta-analyses on influenza virus, herpes virus, HBV, scarlet fever, mumps, chicken pox, pertussis, German measles, and measles virus in this study.
Sensitivity analysis showed that heterogeneity when meta-analyzing influenza virus, herpes virus, measles, and mumps could be completely or partially attributed to Vlajinac's study (Vlajinac et al., 2013). This may be because that Vlajinac's study (Vlajinac et al., 2013) determined infection by questionnaire rather than medical record or laboratory examination, which may induce recall bias and overestimation of viral infection. Interestingly, when we excluded Vlajinac's study (Vlajinac et al., 2013), the meta-analysis found a significantly decrease risk of PD in patients with measles infection, which was in consistent with previous studies (Harris et al., 2012). It is considered that Measles infection in younger age may induce immune response which was speculated to protect the infected person from other viral infection-induced substantia nigra damage and thus decreased the risk of PD (Sasco & Paffenbarger, 1985). However, the specific mechanism of this F I G U R E 2 Forest plot for the pooled odds ratios (ORs) showed positive association between Helicobacter pylori (HP; a), hepatitis C virus (HCV; b), Malassezia (c), pneumoniae (d), and measles (f) the risk of Parkinson's disease (PD). Pooled OR of measles showed significant association with risk of PD after excluding one study that was responsible for heterogeneity (e) Additionally, although significant heterogeneity exists among studies on HCV infection which could be mainly attributed to Golabi's study (Golabi et al., 2017), we failed to figure out the specific reason of heterogeneity.
Infection of other pathogenic microorganisms was also re-

ACK N OWLED G M ENTS
This study is supported by National Natural Science Foundation of China (81771391).

CO N FLI C T O F I NTE R E S T
All authors declare that they have no conflict of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that supports the findings of this study are available in the tables of this article.