The Relationship between Pneumocystis Infection in Animal and Human Hosts, and Climatological and Environmental Air Pollution Factors: A Systematic Review

Background: Over the past decade, there has been rising interest in the interaction of Pneumocystis with the environment. This interest has arisen in part from the demonstration that environmental factors have important effects on the viability and transmission of microbes, including Pneumocystis. Environmental factors include climatological factors such as temperature, humidity, and precipitation, and air pollution factors including carbon monoxide, nitrogen dioxide, sulfur dioxide, and particulate matter. Methods: We undertook a systematic review in order to identify environmental factors associated with Pneumocystis infection or PCP, and their effects on human and animal hosts. Results: The systematic review found evidence of associations between Pneumocystis infection in animal and human hosts, and climatological and air pollution factors. Data from human studies infers that rather than a seasonal association, presentation with PCP appears to be highest when the average temperature is between 10 and 20°C. There was evidence of an association with hospitalization with PCP and ambient air pollution factors, as well as evidence of an effect of air pollution on both systemic and bronchoscopic lavage fluid humoral responses to Pneumocystis. Interpretation of human studies was confounded by possible genetically-determined predisposition to, or protection from infection. Conclusions: This systematic review provides evidence of associations between Pneumocystis infection in both animal and human hosts, and climatological and environmental air pollution factors. This information may lead to an improved understanding of the conditions involved in transmission of Pneumocystis in both animal and human hosts. Such knowledge is critical to efforts aimed at prevention.


Introduction
Pneumocystis jirovecii is a fungus that continues to be an important cause of pneumonia (PCP) in the immunocompromised host and a major cause of death in humans [1]. It is estimated that there are more than 400,000 annual cases of PCP worldwide, with over 52,000 deaths per year [2]. Knowledge of the basic biology of Pneumocystis has long been limited by of the lack of a reliable and reproducible method of in vitro cultivation [3] but important insights have been gained from both animal and human studies. Pneumocystis organisms found in different hosts are morphologically indistinguishable but host speciesspecific [1]. Colonization of apparently healthy asymptomatic humans may provide a reservoir of P. jirovecii, and transmission of Pneumocystis to both susceptible and healthy persons may occur. However, the exact relationship is incompletely understood, and environmental reservoirs for the organism have also been suggested. Pneumocystis infection is acquired by inhalation, and the infective moiety is the cystic form [4] but the precise conditions for airborne transmission are unknown. Traditionally, PCP was thought to be the result of reactivation of latent infection acquired early in childhood, but molecular studies from PCP outbreaks demonstrate that PCP can also result from recent exposure in an at-risk host [5]. As a result, an improved understanding of the conditions involved in transmission is critical to efforts at prevention. undertook this systematic review in order to identify specific environmental factors associated with Pneumocystis infection or PCP, and their effects on human and animal hosts.
Our goal was to comprehensively review the published literature on this topic in order to gain insights and advance our understanding of this important human pathogen.

Materials and Methods
We sought to identify publications describing associations between environmental factors and detection of Pneumocystis in humans and animals, and with development and presentation with PCP. We first reviewed English-language published articles of Pneumocystis and PCP and associated climatological and air pollution factors for the period 01 January 1960 to the present day (30 June 2018) using PubMed (US National Library of Medicine). The following search terms were used: Pneumocystis [Title] + English [Language], then Human [MeSH], then one of the following MeSH: season, climate, air pollution, environment, geography, humidity, or temperature. In addition, we reviewed the references within each publication for additional articles. Since the first reports of human immunodeficiency virus (HIV)-associated PCP in the early 1980s, most cases of PCP in the literature have been described in the context of HIV infection.
The scope of the present study was increased to include animals studied in the wild, in slaughterhouses, and in research laboratories. The same time period was used and the following search terms were used in the literature search: Pneumocystis [Title] + English [Language], then, Animal [MeSH], then one of the following MeSH: season, climate, air pollution, environment, geography, humidity, or temperature. Inclusion of these animal studies was done in order to enhance the findings from human studies and to provide important insights that add to our understanding of studies in humans.
Studies identified by the search strategy were divided into three groups: Group 1: animal studies. Group 2: human studies of HIV-infected and uninfected patients with PCP divided based on geographic location. Group 3: human studies of HIV-infected patients with PCP associated with ambient air pollutants.
Where specific climatological information (i.e., temperature and humidity/precipitation) was not described in a specific publication we used World Weather Online [7]. In addition, as this is a systematic review of previously published work, Ethics Committee/IRB approval was not required from any of the three centers involved in this work.

Literature Search
The PubMed literature search for the period 01 January 1960 through 30 June 2018 identified 74 unique full text articles. From these, 53 articles were excluded (after review of the full text by LH and RFM), as they were not relevant to the topic, or they were review articles that did not contain original data. In addition to the 21 remaining articles, 16 articles were identified by hand searching by the authors (LH and RFM). Additionally, the authors of some publications identified either by the MeSH search or via the "hand search" were contacted (by RFM) in order to obtain additional information not contained in their original published manuscript. Thus, in total 37 articles were identified and were included in the systematic review. Figure 1 shows the results of the literature search.

Animal Studies Showing Associations between Detection of Pneumocystis and Seasonal and Environmental Factors
Group 1. Eleven studies in animals (six from Europe, two from South America, two from Asia, and one from USA) were identified. Animal hosts were pigs (four studies), wild mouse species, shrew species, field voles, rats, hares, crab-eating macaques, and bats (each one study), (Table 1) [8][9][10][11][12][13][14][15][16][17][18]. Pneumocystis was detected in all of the animal species studied, using a range of techniques including Grocott silver staining, immunohistochemistry, and polymerase chain reaction (PCR). Using PCR, Pneumocystis was detected in as few as 4-5% of pigs in a slaughterhouse [17] to as many as 34.5% of crab-eating macaques living in a Primatology Center [12].
The main findings were that in eight of the 11 studies there was a clear seasonal variation in detection of Pneumocystis among different hosts, in their natural surroundings [8,9,10,11,14,15,17,18]. In the study of macaques housed in a Primatology Center, detection was associated with mean precipitation, and in the study of rats in a laboratory facility, temperature and humidity had a clear influence on the predominance of different types of Pneumocystis. Finally, the study of bats in both captive and wild environments showed no association between detection of Pneumocystis and either temperature or humidity, but an association with smaller, crowded sedentary colonies at altitudes below 800m above sea level.

Human Studies Showing Associations between Pneumocystis and Seasonal and Environmental Factors
Group 2. Twenty-four studies in HIV-infected and uninfected patients with PCP were identified and were divided, based on geographic location. Seven studies were from USA, two from South America, 11 from Northern Europe, three from Southern Europe, and one from Australia, (Table 2) [6,. These studies were a mixture of prospective and retrospective clinical or laboratory (including autopsy) studies and were national, multicenter or single center in their design. Two studies were done in infants, the remaining 22 studies were done in adults. Among the adult studies 19 included only HIV-infected adults, two included HIV-infected and uninfected adults, and one included HIV-uninfected adults with underlying rheumatologic conditions. In the two studies of infants, subjects were HIVuninfected .
The main findings were that in 18 studies in adults there was evidence of a seasonal and/or climatological association and presentation with PCP, or detection at autopsy and two studies showed no such association but noted clustering of cases of PCP by Zip Code, and another identified that time spent outdoors was associated with risk of PCP. In the remaining two studies there was no apparent seasonal or climatologic association with development of PCP. However, both studies showed an apparent ethnic predisposition, with patients of black African origin being at reduced risk of developing PCP compared to patients who were of Western origin (Table 2). Both studies in infants showed a seasonal variation either in detection of P. jirovecii at autopsy, or variation in antibody responses to Pneumocystis major surface antigen components.

Human Studies Showing Associations between Pneumocystis and Ambient Air Pollution Factors
Group 3. Four studies (three from USA, one from Spain) reported the impact of ambient air pollution factors among HIV-infected adult patients hospitalized with PCP (Table 3) [6,40,42,43]. Of note two of these four studies are also included in Group 2 [6,42]. The San Francisco studies were single-center, prospective studies while the study from Spain was a national study.
The main findings were that one study showed elevated levels of NO 2 , PM 10 and ozone in ambient air were associated with increased risk of hospitalization with PCP [40], and in another SO 2 , was associated with increased risk of hospitalization, but the risk was attenuated by elevated CO levels [6]. Two further studies that examined serologic responses in hospitalized patients with PCP, and showed elevated NO 2 , and PM 10 were independently associated with impaired IgM responses to P. jirovecii major surface glycoprotein (Msg) constructs in serum [42], and that there was an impaired IgA response to P. jirovecii Msg in bronchoscopic lavage (BAL) fluid that was associated with increased ambient ozone exposure. Additionally, increased BAL fluid IgA responses were associated with increased ambient NO 2 exposure [43] (Table 3).

Discussion
We believe that the present study is the first systematic review of the relationship between Pneumocystis infection in its animal and human hosts, and the effects of climatological and air pollution factors in the environment on this relationship. This review found evidence of associations between Pneumocystis infection in animal and human hosts, and climatological and environmental air pollution factors, but the quality of evidence was poor and was limited by inconsistent and incomplete sampling methodology in both animal and human studies.

Non-Pneumocystis Fungi -Seasonal and Climatological Factors
Climatological factors, including temperature and humidity, are associated with variations in concentrations of ascomycetes, basidiomycetes, and other fungi in air [44][45][46][47][48]. A study from Porto Alegre, Southern Brazil, showed the highest detection of airborne fungal spores was in summer (December-February), when average minimum air temperatures are typically between 18 and 21°C [7]. The lowest rates of detection were in the autumn (March-May), when average minimum temperatures are typically between 12 and 17°C [45]. A second study from Fortaleza, North East Brazil, where the climate is hotter, reported seasonal variation in rates of detection of airborne fungi, (higher rates of detection were observed in January-April, and lower rates were recorded in July-October [46]. This city experiences a rainy season (February-July: rainfall 1883.5 mm, average temperature 26.8°C) and a dry season (August-January: rainfall 260 mm, average temperature 27.6°C) [46].

Pneumocystis -Seasonal Factors -Average Temperature
In the studies of Pneumocystis in animals studied in the wild, in slaughterhouses, and in research laboratories that were identified in this systematic review interpretation of the observed seasonal variation is confounded by the fact that none of the studies systematically sampled throughout the year, nor sampled in consecutive years, and none of the studies included "controls". Thus, possible alternative explanations for the apparent seasonal variation must include closer animal-to-animal proximity within animal colonies, driven by climatological factors, and by season, for example by closer co-habitation during the mating season for an individual animal host. This potentially increases the chance for airborne or "other" animal-to-animal transmission, and thus likelihood of detection by opportunistic, or systematic sampling. In the studies of pigs, another factor influencing detection of Pneumocystis is the type of husbandry used in rearing the animals, as higher rates of detection were observed in semi-intensively farmed animals when compared with pigs who were intensively-farmed. This finding has been ascribed to less control of production variables in pigs reared semi-intensively [17]. Additionally, detection of Pneumocystis has been associated with underlying viral and/or bacterial infection in pigs. It has been suggested that these infections may act as potential "immune suppressants" thus permitting colonization/infection with Pneumocystis [15,17] Taken together the human studies clearly demonstrate evidence of a seasonal and/or climatological association and presentation with PCP, or serologic or autopsy detection of infection. A seasonal variation in incidence of PCP was first reported in 1991 [19]. Subsequently, the relationship between seasonality and temperature and hospitalization with PCP has described inconsistent findings with some, but not all studies describing any association. In both HIV-infected and uninfected patients the risk of PCP has been observed to be higher in summer (London, UK) [28,35], (Geneva, Switzerland [29] (Munich, Germany) [36] and (San Francisco, USA) [6] autumn (Melbourne, Australia) [41], and winter (Seville, Spain) [40]. As previously suggested [6], these differing results might be as a result of differences in patient populations, climatological factors, geography, Pneumocystis genotypes, or to differences in study design [6]. Because of geographical climatological differences, it is likely that summer temperatures in one country, or region, are similar to autumn or winter temperatures in another geographical location. For example, in Spain winters are generally mild (most regions having an average temperature between 10 and 20°C), and summers are hot (with average temperatures greater than 30°C) [40]. Thus, the average temperature in a Spanish winter is similar to that observed in other seasons in other countries during which the highest incidence of PCP has been described [40]. In London the "peak" summer temperature was 13°C *28], the mean summer temperature in San Francisco was 17.6°C [6], and average temperature in Melbourne was between 13.2 and 20°C [7,40]. Taken together, interpretation of these data, infer that rather than a seasonal association, presentation with PCP appears to be highest during the season of the year when the average temperature is between 10 and 20°C [40]. As previously noted [40], among HIVinfected persons, other confounding factors, including tobacco smoking [24], chronic obstructive pulmonary disease, bacterial pneumonia, and colonization of the respiratory tract by P. jirovecii, can contribute to increased risk of developing PCP, and also that these contributing factors additionally, are potentially affected by climatological factors [24].
A similar seasonal association is apparent in the two studies of infants done in Chile [25,26]. In the autopsy study the highest rate of detection of Pneumocystis was in winter (June-August), when average maximum temperatures were 16-18°C, and the lowest rate of detection was in autumn (March-May), when average maximum temperatures were 20-28°C [7,25]. In the study of serum antibody responses to Pneumocystis MsgA constructs among immune competent infants the lowest peaks were also detected in autumn (March-May) [26].

Seasonal Factors -Humidity and Precipitation
There is a less certain association between the climatological variables of humidity and precipitation, and detection of Pneumocystis in animal and human hosts, or with presentation with PCP in humans. Some animal studies [12,17,18] and some human studies [20,27,28,33] suggest an association, but others do not. Confounding these observations is the fact that data about this climatological variable was not routinely collected in either the animal or human studies.
A further potential factor, confounding interpretation of data concerning both temperature and humidity/precipitation in humans is that in the EuroSIDA study, which was a prospective observational cohort study that reported diagnosis of PCP in North, Central, and South Europe, the higher prevalence of PCP described in cooler, wetter North European climates might be explained by other logistical factors, including better access to healthcare (and so the likelihood of being diagnosed with PCP) in North Europe [30]. A similar interpretation can be applied to the findings from another European retrospective multicenter study [31].

Seasonal Factors -Outdoor Activities
The study from Atlanta, USA showed spending time outdoors gardening, camping or hiking in the six months prior to hospitalization with pneumonia was strongly associated with risk of PCP [21]. Additionally, studies from Cincinnati and San Francisco, USA showed clustering of PCP by Zip Codes [22,24]: the former study showing more cases in affluent areas with more green space [22], the latter reporting more cases in residential areas containing parks and small yards [24]. Taken together, these three studies infer that susceptible individuals are at greater risk of developing PCP if they have increased opportunities for being outdoors, and thus possibly for increased environmental exposure to Pneumocystis.

Genetic Factors
Observations from national (Netherlands) and single-site London (UK) cohorts infer that there might be an ethnic/genetically-determined predisposition to development of PCP [32,37]. Single point mutations (SNPs) in genes associated with innate immune function are increasingly recognized as significant factors dictating an individual's susceptibility to infection *49 -51]. Two reports describe an association between SNPs and development of PCP in HIV-infected individuals [52,53]. The FcγIIa receptor, which binds to immune complexes to facilitate uptake of microbes, is encoded by a gene with synonymous (functional) polymorphisms that affects binding affinity to IgG. Participants in the Multicenter AIDS Cohort Study (MACS) with the FcγRIIa RR genotype progressed to AIDS (i.e., CD4 count <200 cells/uL) faster than participants with RH or HH genotypes [52]. By contrast, participants with an FcγRIIa HH genotype progressed more quickly to AIDS (defined by development of PCP) than those with other genotypes [52]. While the underpinning mechanisms remain uncertain, these results raise intriguing questions about the antibodies that opsonize Pneumocystis and interact with FcγRIIa receptors. The chemokine receptor CXCR6 is a co-receptor that facilitates fusion of HIV to CD4 cells. The ligand for CXCR6 (CCXCL16) is highly expressed in the lungs. A SNP in codon 3 (CXCR6-E3K) is common in African Americans and rare in Caucasians. The AIDS Link to Intravenous Experience (ALIVE) study, a prospective cohort study of predominantly African American HIV-infected drug users, examined the relationship of CXCR6 SNPs and development of PCP [53]. Time to development of PCP was similar among the genotypes, but subjects homozygous or heterozygous for CXCR6-3E were more likely to die after PCP (and thus had a shorter survival time) than subjects homozygous for CXCR6-3K [53].
The current epidemiology of pediatric PCP in the USA demonstrates that it is less frequently observed in HIV-infected children. By contrast cases associated with hematologic malignancy and primary immunodeficiency have become more prominent, infants being the most commonly affected [54]. Well-described mutations in MHC class II (bare lymphocyte syndrome), recombination activating genes (RAG) −1 and −2, signal transducer and activator of transcription (STAT) −3, and IL-21R that can be modeled in genetically engineered mice and P. murina infection, provide evidence of genetic susceptibility to Pneumocystis infection [55]. Taken together these data suggest a gene-environment interaction in the disease.

Air Pollution
Among the general population it is increasingly evident that ambient air pollution contributes to the global burden of respiratory disease, including asthma, Chronic Obstructive Pulmonary Disease, and pneumonia [56][57][58][59][60][61][62][63][64][65]. The studies identified in this systematic review originating from Spain and San Francisco, USA clearly demonstrate an association between ambient air pollution factors and presentation of HIV-infected adults with PCP [6,40]. It is intriguing that in one study there was no association with PM 10 , NO 2 , CO, or ozone, and an association was only evident for SO 2 [6]. In the other study NO 2 , PM 10 , CO, and ozone were associated with risk of hospitalization with PCP [40]. This apparent difference may in part be explained by the fact that San Francisco is one of the least polluted cities in the United States and so it is possible that levels of these pollutants were below thresholds that can trigger respiratory complications [6].
The mechanisms by which ambient air pollution increases susceptibility to pulmonary infection are not well defined. Controlled exposure studies of single pollutants that have used cells, animals and human subjects indicate that ambient air pollutants alter innate lung immunity at multiple levels, including altered muco-ciliary function, respiratory epithelial cell dysfunction, impaired alveolar macrophage phagocytosis, and dysfunction of surfactant protein A and D [Reviewed in 6]. However, the effects of ambient air pollution factors on humoral immunity and serologic responses to pulmonary infection, and the immune-toxic effects of real-life exposures to ambient air pollution factors are poorly understood. In one study from San Francisco [42] PM 10 and NO 2 were independently associated with suppressed IgM (but not IgG) responses to Pneumocystis Msg constructs. It was suggested that PM 10 particles encountering bronchus associated lymphoid tissue, might impair antigen presenting cell function, resulting in decreased activation of the humoral immune system and suppressed serologic responses [42]. Both animal and human exposure studies have found mixed effects of NO 2 inhalation on bronchoalveolar and systemic antibody responses [66][67][68][69]. In a second study also from San Francisco patients with PCP increasing exposure to ozone was associated with reduced BAL fluid IgA responses to P. jirovecii Msg constructs and increasing exposure to NO 2 was independently associated with increased BAL fluid IgA responses to P. jirovecii Msg [43]. These results might be explained by the observation that ozone is a potent oxidant resulting in both bronchoalveolar, and systemic inflammation. Animal studies have found that rats in the first two weeks of exposure to ozone demonstrate decreased antibody responses to microbial antigens such as Listeria spp [70].

Conclusions
This systematic review found evidence of associations between Pneumocystis infection in both animal and human hosts, and climatological and environmental air pollution factors. These data are limited by inconsistent and incomplete sampling methodology in both animal and human studies. Data from human studies infer that rather than a seasonal association, presentation with PCP appears to be highest when the average temperature is between 10 and 20°C. A potential confounder is possible genetically-determined predisposition to, or protection from infection. There is evidence of an association with hospitalization with PCP and ambient air pollution factors, as well as a   Table 1 Animal studies showing associations between detection of Pneumocystis spp and seasonal and environmental factors.

Study authors
Year Species Location

Main findings
Šebek & Rosický [8] 1967  Bacchetti [20] 1994 USA Analysis of monthly trends in AIDS diagnoses among adolescents and adults reported to CDC. Seasonal pattern of PCP, with a peak in March. More PCP in MSM/bisexual men in summer in West of USA (includes San Francisco and Los Angeles) which is drier than rest of USA at this time.
Navin, et al [21] 2000 Atlanta, Georgia Two centre single city retrospective cohort study. PCP associated with outdoor activities (gardening, hiking or camping).
Dohn, et al [22] 2000 Cincinnati, Ohio Single centre cohort study. Clustering of PCP cases by Zip code. Most cases occurred in affluent areas with more green space. No seasonal variation.
Morris, et al [23] 2000 San Francisco, California Single centre cohort study. Clustering of PCP cases by Zip code. Most cases occurred in more residential areas that contain parks and small yards. No seasonal variation.
Morris, et al [24] 2004 MACS cohort Four US cities Retrospective cohort study of patients who had necropsy. PCR detection of P. jirovecii.
Rates of colonization (determined by PCR) greater in cigarette smokers. Detection rates higher in Chicago and Pittsburgh (70.4% and 61.5%) than in Baltimore and Los Angeles (42.3% and 16%).
Djawe, et al [6] 2013 San Francisco, California Single centre cohort study. Seasonal variation in hospitalization with PCP; peak in summer (June-August).

South America
Vargas, et al [25] 2005 Santiago, Chile Autopsy study, two children's hospitals. PCR detection of P. jirovecii. P. jirovecii DNA detected in 51.7% of immune competent infants dying in the community and in 20% dying in hospital. Seasonal variation in detection: 53% in Winter (June -August) when maximum temperature = 16-18°C. and 30% in Autumn (March -May) when maximum temperature =20-28°C.