The prevalence of microsporidia in China : A systematic review and meta-analysis

Microsporidia are a diverse parasite phylum infecting host from all major taxa in all global biomes. This research was conducted to conclude the prevalence of microsporidia in China. All published articles up to February 16, 2018 were considered, including descriptive, cross-sectional, case-control and epidemiology studies. A total of 1052 articles were separated after literature search. After a strict selection according to our criteria, 82 articles were included in qualitative synthesis and ultimately 52 studies were included in quantitative synthesis. Three species of microsporidia were confirmed to exist in China, including Enterocytozoon bieneusi (E. bieneusi), Nosema and Encephalitozoon cuniculi (E. cuniculi). The highest overall estimated prevalence of E. bieneusi in humans was 8.1%, which was observed in acquired immunodeficiency syndrome patients (AIDS). Moreover, the prevalence of E. bieneusi in animals including the cattle, dogs, pigs, deer, sheep and goats were analyszed in this study. The overall estimated prevalence of E. bieneusi acquired by using the random effects model in meta-analysis in cattle, dogs, pigs, sheep and goats and deer was 20.0% (95% confidence intervals: 0.133–0.266, I2 = 98.031%, p < 0.0001), 7.8% (95% CI: 0.050–0.106, I2 = 60.822%, p = 0.0537), 45.1% (95% CI: 0.227–0.674, I2 = 98.183%, p < 0.0001), 28.1% (95% CI: 0.146–0.415, I2 = 98.716%, p < 0.0001) and 19.3% (95% CI: 0.084–0.303, I2 = 96.995%, p < 0.0001) respectively. The overall detection rate of E. bieneusi in water acquired by using the random effects model in meta-analysis was 64.5% (95% CI: 0.433–0.857, I2 = 98.486%, p < 0.0001). Currently, 221 genotypes of E. bieneusi, 1 genotype of E. cuniculi and 6 Nosema were detected in China. The most prevalent genotype of E. bieneusi was genotype D, followed by BEB6 and EbpC.

www.nature.com/scientificreports www.nature.com/scientificreports/ Trachipleistophora hominis and Encephalitozoon species (E. cuniculi, Encephalitozoon hellem, and Encephalitozoon intestinalis) were associated with keratoconjonctivitis 5 . In terms of therapy, albendazole is effective against E. intestinalis but not on infection with E. bieneusi. Though Fumagillin has shown the clinical therapeutic effect on E. bieneusi 5 , its efficacy is counterbalanced by its adverse effects 5 . Over the last decade, PCR amplification and staining techniques have been the most common approaches for detection, when sequence analysis of the internal transcribed spacer (ITS) has been widely used in characterizing the molecular epidemiology 9 . To this end, this study was carried out a systematic review with meta-analysis of microsporidia studies in vertebrate and invertebrate hosts and Chinese population distributed in different regions in China. The present study may be the first meta-analysis that provides overall results based on available molecular and staining methods. According to this systematic review, not only we can improve awareness about microsporidia prevalence in various regions of China, but we will also be able to implement better preventive and treatment strategies.

Results
A total of 1050 articles were separated after literature search of 8 databases (Pubmed: 19, Embase: 60, Web of Science: 151, Cochrane library: 0, CNKI: 403, Wanfang: 289, VIP: 128, CBM: 0), with 2 articles coming from other sources, and ultimately 82 articles were included in qualitative synthesis (Fig. 1). After excluding 6 studies with controversial data and 24 studies with essential data deficiency, 52 studies were included in quantitative synthesis (meta-analysis). All data extracted from included articles was demonstrated in Table 1 and Supplementary  Table S1 to S4. As few articles have reported the prevalence of Nosema 10,11 and Encephalitoznoon spp. [12][13][14] , it is not clear about their real prevalence. Thus, our further analysis only emphasised on E. bieneusi. Moreover, our statistical analysis of E. bieneusi was confined to human-beings, water, cattle, dogs, pigs, deer and sheep and goats.

Discussion
Our aim was to conclude the prevalence of microsporidia in China, and according to our statistics, higher prevalence was observed in animals and water when compared with that in human-beings. For example, the highest infection rate reported in humans was 22.2%, while that was 100% in both animals and water. The poor living condition may be responsible for the high prevalence of animals. As for the high prevalence in water, there were a total of ten records on water, and six of these records focused on waste water. Waste water is a general term for water and runoff rainwater discharged during residents' activities, which is seriously polluted by feces, domestic garbage and industrial waste. The high prevalence of E. bieneusi in water may be attributed to the pollution. Moreover, it was reported that water, either consumed directly by drinking or indirectly via irrigating or washing foods, bathing, washing, or for recreation, provided a crucial medium for spore survival and transmission 19 .
Furthermore, our data showed the highest overall estimated prevalence of E. bieneusi in humans was observed in AIDS patients (8.1%). In nature, there is a balanced interaction between E. bieneusi and their hosts, which leads to long-term subclinical infections 20 . When the host is immunocompromised, infection can cause overt signs of clinical disease 21 . It was reported that AIDS patients whose CD4 + T cells counting was less than 50 per mm 3 blood were most likely to experience persistent diarrhea, weight loss, and abdominal pain associated with E. bieneusi or E. intestinalis infections. Since the first case of AIDS was reported in 1985, the prevalence of AIDS in China has become increasingly severe. According to a joint assessment by the Chinese Center for Disease  www.nature.com/scientificreports www.nature.com/scientificreports/ Control and Prevention, Joint United Nations Programme on HIV and AIDS and the World Health Organization, by the end of 2018, there will be approximately 1.25 million people living with HIV in China. Such a large number makes us think about how to improve their life quality. Thus, further studies on therapy and prevention are needed as there is no effective solution currently. At the same time, the high prevalence of E. bieneusi in diarrhea patients should not be ignored as there are approximately 70 million cases suffering from diarrheal diseases annually in China. On the other hand, considering the high prevalence, we suggest clinical doctors taking E. bieneusi infection into consideration when dealing with diarrheal issue and AIDS patients. Also, for laboratory researchers who need E. bieneusi strain for further study, feces from AIDS patients and diarrheal patients may be the first choice for separating E. bieneusi strains.
To our knowledge, this is the first systematic review and meta-analysis of the prevalence of E. bieneusi in China. However, there is some limitation of this meta-analysis which may influence the results. Firstly, the research on human-beings was insufficient as there were only six provinces, one autonomous region and one municipality have reported the infection of humans. Research on humans still has a lot of blanks waiting for us to fill. Secondly, in order to fully understand the prevalence of E. bieneusi in animals, more different kinds of animal hosts should be included. Moreover, instead of being confined to several provinces, sampling regions of the same host need to be expanded. Thirdly, repeat detection and negative control were ignored in some studies, which may bring about an inaccurate result. Fourthly, included studies on water mainly focused on waste water, which prevented us from further analysis of other water source, such as combined sewer overflow and drink water. Fifthly, the lack of usable data forced us to give up the analysis of other species of microsporidia and focused only on E. bieneusi.
In conclusion, this review provides a broad outlook of the prevalence of E. bieneusi in China, but there are still some problems that need to be solved by all of us. In addition, as chemoprophylaxis and chemotherapeutic treatment modalities are not available, virtually nothing is known about immunity, vaccines are nonexistent, and the study of E. bieneusi are supposed to be more further improved. Meanwhile, it's necessary to strengthen the prevention. Effective measurements including protection of water source and food from pollution, environmental hygiene, personal hygiene and health education should be emphasized.

Materials and Methods
The current study followed the Meta-analysis Of Observational Studies in Epidemiology (MOOSE) guidelines 22 (Supplementary Data Table 5).

Search strategy. The article searching was conducted in 8 databases, including both English articles and
Chinese articles, and 2 articles came from other sources. Pubmed, Embase, Web of science and Cochrane library were applied for looking up English articles, while Chinese articles searching was conducted in CBM, CNKI, VIP and Wanfang databases. Manual searches were proceeded as a supplement. Time limitation of publication was no later than February 16, 2018. Keywords applied for searching were Microsporidiosis, Microsporidia, Microsporidium, Microspora, Nosema, E. bieneusi, Encephalitozoon spp., China, Epidemiology, Genotype and Prevalence.

Study selection.
Articles had to meet the following inclusion criteria: all chosen articles should be published up to February 2018; the study should be one of the following types: descriptive study, cross-sectional study, case-control study or epidemiology study; the language of the article was supposed to be either English or Chinese; the studies should be closely related to the prevalence and genotypes of microsporidia.
Articles were excluded if they meet any of the following exclusion criteria: studies that used other diagnostic methods, except staining and molecular techniques; articles written in a language other than English and Chinese; unscientific publication about microsporidia infection (abstracts, national conference proceedings); duplicate studies with overlapping data; articles had poor association with the prevalence and genotypes of microsporidia.
The suitability of all studies was considered by three different authors (L.Q., W.L. and J.P.). Discrepancies were resolved by consensus. After selecting articles, the following information was recorded in a standard data extraction form. A flow diagram of the study design process has been shown in Fig. 1. Methodological Quality. The methodological quality of the included studies with an accessible full text was independently assessed by 2 reviewers (W.L. and J.P.) using a modified Newcastle-Ottawa scoring guide 23 . To assess the quality of included studies, the scoring approach was applied. One point each were given to the studies if they comply with the following scoring guidelines: I Population contained a mixture of specialties at multiple sites; II Large sample sizes (no less than 200); III Comparability between respondent and non-respondent characteristics was established, and the response rate was satisfactory; IV Repeat detection; V Reported descriptive statistics to describe the population (e.g., age, sex) with proper measures of dispersion (e.g., standard deviation, standard error, range). Up to five points could be assigned to each study (low quality: 0~2 points; moderate quality: 3 points; high quality: 4~5 points). Discrepancies were resolved by consensus or arbitrated by a third reviewer (W.X.). The results were demonstrated in Tables 1, 2.
Data extraction and analysis. After study selection, the following data were extracted by three authors (L.Q., W.X. and W.L.) independently from included articles: hosts, province, year of publication, total participants, positive cases, detection method, genus, gender and immune function status. Considering the close relationship between E. bieneusi and HIV/AIDS patients with diarrhea, which is the most typical symptom, individuals were divided into four groups: including diarrhea patients, HIV/AIDS patients, other patients and healthy people. Patients with both diarrhea and HIV positive were divided into HIV/AIDS group. The group of other www.nature.com/scientificreports www.nature.com/scientificreports/ patients was composed of cancer patients, outpatients and in-hospital patients. Healthy people group included control group of case-control studies and students. Fact estimates and 95% CI of prevalence of all involved studies were assessed. The random-effects model and subgroup analysis were conducted to investigate the potential source of high heterogeneity. The total prevalence and group-specific prevalence were considered by gender (male and female), sampling region and stool appearance. Forest plots were used to show the heterogeneity among the studies, which showed proportions of individual studies and total prevalence. The meta-analysis was completed with R3.4.3. Distribution map of animal infection rate was drawn with Photo-shop, and the maps of humans and water were omitted for the lack of data.