The largest meta-analysis on the global prevalence of microsporidia in mammals, avian and water provides insights into the epidemic features of these ubiquitous pathogens

Background Microsporidia are obligate intracellular parasites that can infect nearly all invertebrates and vertebrates, posing a threat to public health and causing large economic losses to animal industries such as those of honeybees, silkworms and shrimp. However, the global epidemiology of these pathogens is far from illuminated. Methods Publications on microsporidian infections were obtained from PubMed, Science Direct and Web of Science and filtered according to the Newcastle-Ottawa Quality Assessment Scale. Infection data about pathogens, hosts, geography and sampling dates were manually retrieved from the publications and screened for high quality. Prevalence rates and risk factors for different pathogens and hosts were analyzed by conducting a meta-analysis. The geographic distribution and seasonal prevalence of microsporidian infections were drawn and summarized according to sampling locations and date, respectively. Results Altogether, 287 out of 4129 publications up to 31 January 2020 were obtained and met the requirements, from which 385 epidemiological data records were retrieved and effective. The overall prevalence rates in humans, pigs, dogs, cats, cattle, sheep, nonhuman primates and fowl were 10.2% [2429/30,354; 95% confidence interval (CI) 9.2–11.2%], 39.3% (2709/5105; 95% CI 28.5–50.1%), 8.8% (228/2890; 95% CI 5.1–10.1%), 8.1% (112/1226; 95% CI 5.5–10.8%), 16.6% (2216/12,175; 95% CI 13.5–19.8%), 24.9% (1142/5967; 95% CI 18.6–31.1%), 18.5% (1388/7009; 95% CI 13.1–23.8%) and 7.8% (725/9243; 95% CI 6.4–9.2%), respectively. The higher prevalence in pigs suggests that routine detection of microsporidia in animals should be given more attention, considering their potential roles in zoonotic disease. The highest rate was detected in water, 58.5% (869/1351; 95% CI 41.6–75.5%), indicating that water is an important source of infections. Univariate regression analysis showed that CD4+ T cell counts and the living environment are significant risk factors for humans and nonhuman primates, respectively. Geographically, microsporidia have been widely found in 92 countries, among which Northern Europe and South Africa have the highest prevalence. In terms of seasonality, the most prevalent taxa, Enterocytozoon bieneusi and Encephalitozoon, display different prevalence trends, but no significant difference between seasons was observed. In addition to having a high prevalence, microsporidia are extremely divergent because 728 genotypes have been identified in 7 species. Although less investigated, microsporidia coinfections are more common with human immunodeficiency virus and Cryptosporidium than with other pathogens. Conclusions This study provides the largest-scale meta-analysis to date on microsporidia prevalence in mammals, birds and water worldwide. The results suggest that microsporidia are highly divergent, widespread and prevalent in some animals and water and should be further investigated to better understand their epidemic features. Supplementary Information The online version contains supplementary material available at 10.1186/s13071-021-04700-x.


Introduction
Microsporidia are a group of ubiquitous and obligate intracellular pathogens [1][2][3]. Over 200 genera and 1400 species of microsporidia have been identified [4]. These pathogens have been widely reported to infect economically important insects, fish, crustaceans, mammals and birds [5][6][7][8][9][10]. Moreover, 17 species have been found to infect humans and cause microsporidiosis [3,4]. Enterocytozoon bieneusi and Encephalitozoon, such as Encephalitozoon cuniculi, Encephalitozoon intestinalis and Encephalitozoon hellem, are the major species identified that infect humans, among which E. bieneusi is the most clinically reported [11,12]. The clinical manifestations of human microsporidiosis are enteritis, cholecystitis and diffuse infection without specific symptoms [13]. Microsporidia can also cause self-limiting infections in immunocompetent individuals and life-threatening chronic diarrhea in immunocompromised populations [14]. Both immunocompetent and immunocompromised individuals run a risk of corneal infection, leading to self-limiting mild keratoconjunctivitis and even severe interstitial keratitis, which is difficult to treat with drugs [15,16]. The infection rate of E. bieneusi in children < 2 years of age has been reported to be 13% in Nigeria [17], 17.4% in Uganda [18] and 11.83% in China [19]. In Australia, fecal samples from children < 3 years of age showed a higher infection rate (2.5%) than those from adults (0.3%) [19]. In addition, advanced age is also a potential risk factor. A study investigated 382 randomly selected people aged 1 to 84 years and showed that the infection rate in people > 50 (56.25%) was much higher than that in adults (38.55%) [16]. Another study surveyed E. bieneusi infection in 60 HIV-negative elderly patients and found that 8 were positive (17.02%) [20], which is higher than the overall rate of 11.8% in HIV-infected people [21]. Immunosuppressive therapy for organ and bone marrow transplant patients could lead to cellular immunodeficiency, which puts them at a high risk for microsporidian infection. In Poland, 11 out of 72 immunosuppressed renal transplant recipients were found to be infected by E. bieneusi [22]. To date, microsporidian infections have been observed in a wide range of human populations, including autoimmune diseases, end-stage renal failure, human immunodeficiency virus (HIV)-positive individuals, leukemia patients and travelers [12,23]. In addition, studies have shown that there is no significant difference in microsporidia prevalence between genders [24].
Microsporidia seem to be ubiquitous and highly divergent in various naturally infected vertebrates [25]. Analysis of ribosomal ITS sequences revealed that some genotypes are present in both humans and animals, posing a public health threat [26][27][28]. Moreover, microsporidia have been detected in a variety of water sources, including irrigation water for crops, recreational water and wastewater from sewage treatment plants [17]. Studies have shown that the overall detection rate of E. bieneusi in water is 64.5% in China [29][30][31]. Researchers speculate that water is a possible container of microsporidia and provides a habitat for spores [32]. Because the chitin-containing spore wall provides protection against various environmental conditions and allows pathogens to survive for long periods, microsporidian spores from symptomatic and asymptomatic hosts could be the source of transmission in humans and animals [27,32,33]. Widespread microsporidia in animal hosts and water cause an important potential risk of human microsporidiosis. Therefore, understanding the epidemiology of microsporidia in animals and water is vital for developing effective measures to prevent the spread and infection of these pathogens. Herein, we conducted a systematic meta-analysis to assess the global prevalence of microsporidia.

Data processing
The included publications were required to investigate the prevalence of microsporidian infections. Data were excluded if they were from repeated studies and reviews, if there was no sample information or if the sample size was < 20, or if they were not determined with staining and molecular techniques. The suitability of all studies was assessed by four different authors. Disagreements were resolved by discussion among the authors.
We assessed the methodological quality of the included studies with an accessible full text according to the Newcastle-Ottawa Quality Assessment Scale [34]. One received a point if the study satisfied the following scoring guidelines: sample collection was random; sample size was > 200; reporting descriptive statistics to describe the population with proper measures of dispersion; reporting results without selectivity; repeating the detection using different methods. Up to five points could be assigned to a study. Publications with a total score of four or five points were regarded as high quality, whereas three points represented moderate quality and lower scores indicated low quality. Studies with a score of less than one point were excluded. After processing, the following data were extracted: country, sampling date, host, number of samples, number of positive samples, genus and species of the pathogen, age, gender and geographic region, and others are listed in Additional file 1: Tables S1-S9. In addition, information about microsporidian species, strains, genotypes, geographic locations and hosts was retrieved from the GenBank nucleotide database (Additional file 2).

Data analysis
Meta-analysis was conducted using Stata version 15.0 to calculate the overall prevalence of microsporidian infections. The chi-squared test-based Q and I 2 statistics were used to estimate the heterogeneity (I 2 < 25%: low heterogeneity; 25% < I 2 < 50%: moderate heterogeneity: I 2 > 50%, high heterogeneity), which presents the percentage of variation between studies. A fixed effect model was used when heterogeneity was < 50%, and a random effects model was used when heterogeneity was > 50%. Due to the high heterogeneity (I 2 > 50%, P < 0.1) in our study, random effects models were used for summary statistics. A forest plot was used to show proportions of individual studies and the total prevalence.
A potential source of heterogeneity was investigated by subgroup analysis and meta-regression analysis. The total prevalence and group-specific prevalence were considered among ages by comparing individuals < 18 years old and > 18 years old, genders by comparing males and females, geographical regions by comparing sub-Saharan Africa with other regions, income levels by comparing low-income countries with countries of other income levels and physical conditions by comparing individuals with HIV and other physical conditions. We also investigated the relationship between CD4+ T cell counts and diarrhea symptoms in the human host. For pig hosts, factors included age group by comparing post-weaned pigs with other ages and species group by comparing pigs with Tibetan pigs and wild boars. For cats and dogs, feral and domestic animals were used to compare living conditions. For cattle and sheep, species comparisons were conducted by comparing yaks and other species and sheep with goats. For nonhuman primates (NHPs), wild and domestic living environments were compared. For birds, factors included bird species by comparing water birds with terrestrial birds and living conditions by comparing wild and domestic living environments. We examined factors both individually and in multiple-variable models. Statistical techniques, P values and coefficients (95% CIs) were used to show the differences in factors.
We analyzed data according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement [35], shown in Additional file 1: Table S10.

Data content
In total, we searched 4129 studies and obtained 287 papers meeting the requirements (Fig. 1), from which 385 epidemiological data records were retrieved (Additional file 1: Tables S1-S9). As the detailed prevalence data were predominantly from E. bieneusi, E. cuniculi, E. hellem and E. intestinalis, our subsequent meta-analysis mainly focused on these four species (Fig. 4b, c).

Global prevalence features of microsporidia Prevalence of microsporidia and coinfection in humans
A total of 92 reports on human infections in 40 countries were retrieved, with 63 reports on E. bieneusi and 14 reports on Encephalitozoon. Regarding the sampling sources, 61 were HIV-positive patients, 22 were immunocompetent individuals, 7 were cancer patients, 7 were other patients and 5 were organ transplant individuals (Additional file 1: Table S1).

Prevalence of microsporidia in other mammals
The prevalence data of microsporidia in other mammals, such as rodents, foxes, raccoons, kangaroos, minks, takins and giant pandas, are shown in Additional file 1: Table S7. Determined with the random effects model in the meta-analysis, the overall prevalence rate in rodents was 17.6% (489/2870; 95% CI 11.6-23.7%) (Additional file 1: Figure S24). In detail, the prevalence rates in rabbits and pandas were 10.2-93% and 6-93%, respectively. In other mammalian populations, however, the overall prevalence could not be estimated because there were insufficient comparable investigations available for meta-analysis.

Geography of microsporidian infections
Data obtained from the GenBank nucleotide database showed that microsporidia are prevalent in 92 countries and regions, where pathogens have been mostly reported in China, Thailand, Russia and India, while investigations in Syria, Switzerland and Romania have been much less prevalent. In addition, microsporidia can infect at least 702 hosts. Among all microsporidia, E. bieneusi is the most widespread species found in both humans and animals and has been detected in 42 countries and mainly reported in Poland, the USA and China. In China, E. bieneusi infections have been actively investigated and found in 148 hosts. In addition, Nosema is the second most reported and has been widely found in silkworms, wasps, mosquitoes and many other animals distributed in 42 countries, including Russia, Japan and Poland. Moreover, Encephalitozoon is widely found in individuals from Rwanda, Australia, Japan and 14 other countries (Additional file 2: Table S11). Among all countries investigated, Turkey, Malawi and Slovakia have the highest prevalence, while Western European countries, such as France, Russia and Italy, have a much lower prevalence (Fig. 5).

Seasonal prevalence of microsporidia
When calculating the seasonal prevalence rate of microsporidia, we found that E. bieneusi was higher in autumn and lower in summer, while Encephalitozoon was more prevalent in summer and less prevalent in winter (Fig. 4d). However, the prevalence rates between seasons showed no significant difference (P > 0.05).

Microsporidian genotypes
Currently, 722 microsporidian genotypes in 7 species have been identified, including 685 E. bieneusi, 14 Loma salmonae, 10 E. hellem, 8 Loma sp. SVB-PE3, 4 E. cuniculi, 3 E. intestinalis and 2 Anncaliia algerae, respectively. To date, the largest number of genotypes has been identified in E. bieneusi, among which genotype D has been the most commonly reported and has been found in 63 hosts, including human and domestic animals in 25 countries (Additional file 2: Table S11). The second largest E. bieneusi genotype, EbpC, has been found in 12 countries. Except for D and EbpC, some E. bieneusi genotypes are rare and have been reported only in one region. For example, HIN1 was only found in Nigeria [27]. Most microsporidia, such as Nosema, Vairimorpha, Vittaforma and Paranucleospora, lack genotype identification and need further study on genetic diversity.
Microsporidian infections in humans and animals have been reported worldwide in 40 and 32 countries, respectively (Additional file 2). However, surveys in water were only conducted in five countries. Considering the high prevalence rate in water, microsporidia investigations should be conducted in more water sources and locations. In addition, detection in wild animals is limited; for example, infectious data are lacking in elephants, peacocks, zebras, koalas and many other wild animals.
Microsporidian prevalence was reported to be related to sanitation facilities, drinking water, animal exposure and diagnostic methods [21,29]. The high prevalence in Northern Europe and South Africa may be related to the developed logistics communication in Northern Europe and underdeveloped health facilities in South Africa (Fig. 5). Furthermore, the prevalence varies greatly in different areas of a country. In Nigeria, for instance, microsporidian infection in humans was 23.3% in Los Lagos [41] but only 7.5% in Ibadan [42]. Therefore, it is necessary to survey and compare the regional distributions in each country.
Notably, microsporidia are highly prevalent in water, posing a high risk for human and animal infection. Microsporidia have been detected in treated effluent and raw sewage [43]. In addition, the same genotypes were detected in wastewater from different treatment plants [30,44,45]. These findings suggested that water contamination was likely impacted by humans, livestock and rodents. Therefore, microsporidia from human and animal excretions entering the environment via sewage wastewater probably led to expansion of infection [21]. Because water is likely an important source of infection, guidelines on wastewater usage are needed to minimize human exposure to microsporidia. It is also necessary to strengthen the detection and disinfection of domestic water.
In addition to water, microsporidia can also be transmitted via food and air. Microsporidia have also been detected in fresh vegetables, fruits and milk [46]. The acceleration of food globalization and transportation could increase parasite transmission. Food-born microsporidia should receive increased attention [47]. Microsporidian spores are also present in air atomized from animal excrement, such as bird droppings, and could be an airborne pathogen [48]. Multiple microsporidian species have been detected in bird droppings [49,50]. Our analysis also demonstrated that E. bieneusi, E. cuniculi, E. hellem and E. intestinalis have been widely found in birds. This implies that the dissemination of airborne microsporidia poses a risk of infection to humans.
Domestic animals showed higher infection rates and are another source of human microsporidiosis. The domestic pig, for example, showed the highest prevalence rate (Fig. 4c). In addition, some superior zoonotic genotypes, such as the D of E. bieneusi, have been widely identified in domestic cats, donkeys, cattle and pigs [51][52][53][54]. Indoor breeding and daily contact with these animals would increase the risk of zoonotic transmission.
The coinfections of microsporidia with other pathogens have been confirmed and should be considered an important public health problem. We found that the coinfection rates of microsporidia with HIV and Cryptosporidium were higher than those of microsporidia with other pathogens. Since 1985, the global AIDS pandemic has been a serious problem [55]. Infections by microsporidia and Cryptosporidium have been frequently reported in HIV-positive patients. Few studies, however, have examined the coinfections of microsporidia with other parasites.
In summary, this study systematically characterized the global prevalence of microsporidia, providing references for future epidemiological studies and pathogen control. However, more periodical surveys are needed to better understand the global and local epidemiological features of microsporidian infections.

Supplementary Information
The online version contains supplementary material available at https:// doi. org/ 10. 1186/ s13071-021-04700-x.    Table S1. Included studies of microsporidian infection in humans. Table S2. Included studies of microsporidian coinfection in humans. Table S3. Included studies of microsporidian infection in swine. Table S4. Included studies of microsporidian infection in cats and dogs. Table S5. Included studies of microsporidian infection in ruminants. Table S6. Included studies of microsporidian infection in nonhuman primates. Table S7. Included studies of microsporidian infection in other mammals. Table S8. Included studies of microsporidian infection in birds. Table S9. Included studies of microsporidia in water. Table S10. Checklist of items included when reporting a meta-analysis.
Additional file 2: Table S11. Information on microsporidian species, strains, genotypes, geographic locations and hosts was retrieved from the GenBank nucleotide database.