Seroprevalence Estimates of Latent and Acute Toxoplasma Infections in HIV+ People—Call for Action in Underprivileged Communities

We undertook a comprehensive, systematic review of observational studies to estimate respective seroprevalences of latent and acute Toxoplasma gondii infections in HIV+ people at the global, regional and country levels; related seroprevalence to socio-economic variables and CD4+ cell counts; and assessed temporal changes in prevalence and risk factors for this group. We systematically searched international databases for seroepidemiological surveys between 1 January 1980 and 31 July 2020. We used a random effects model to calculate pooled seroprevalences with 95% confidence intervals (CI), and estimated the numbers of HIV+ people inferred to harbour latent and acute T. gondii infections (LT or AT). We grouped seroprevalence data according to the geographic regions defined by the World Health Organization (WHO) and conducted subgroup and meta-regression analyses of the data. Of a total of 4024 studies identified, 150 and 65 of them met the inclusion criteria for LT and AT in HIV+ people, respectively. The overall, pooled seroprevalences of LT and AT were 37.4% (95% CI, 33.4–41.4) and 1.3% (95% CI, 0.9–1.8%), equating to ~14.2 and 0.5 million HIV+ people, respectively. Most HIV+ people with T. gondii infections originated from Africa, and the highest seroprevalences were in low-income countries with low human development indices. Significant risk factors for toxoplasmosis in HIV+ patients included the consumption of raw/undercooked meat, frequent contact with soil, a low CD4+ T lymphocyte number (<200 cells per μL) and age. Overall, the finding of high seroprevalences of particularly latent T. gondii infection in HIV+ people in underprivileged regions of the world, such as parts of Africa, calls for preventative action. Programs that include routine serological monitoring, counselling, care, animal control and/or prophylactic treatment measures are needed to prevent severe toxoplasmosis from developing in people living with HIV infection. Our study highlights the potential importance of parasite chemoprophylaxis in resource-poor settings, particularly in low-income countries.


Introduction
Toxoplasma gondii infection can lead to life-threating toxoplasmosis in human immunodeficiency virus (HIV)-infected people [1]. In HIV + people, 'toxoplasmic encephalitis' is a common complication, particularly in individuals with CD4+ T lymphocyte counts of <200 cells per µL [2,3]. This form of toxoplasmosis can be associated with headache, drowsiness, disorientation, hemiparesis, retinochoroiditis and/or epilepsy [4,5], and can become generalised and fatal.
Primary T. gondii infection is acquired through the ingestion of sporulated oocysts in soil or water contaminated with cat faeces, or on contaminated fruits and vegetables; the ingestion of infective stages ('zoites) within raw or undercooked meats; by vertical transmission from an infected mother to the foetus; and, in rare cases, by blood transfusion or organ transplantation [6][7][8][9][10]. While primary T. gondii infection is mostly asymptomatic in immunocompetent people [11], such an infection could lead to severe toxoplasmosis in HIV + people due to their immunodeficiency [12]. More often, however, toxoplasmosis appears to result from the reactivation of a latent T. gondii infection [5,13,14]. Although a low CD4+ T cell count (<200 per µL) is a risk factor for opportunistic infections in HIV + people, severe toxoplasmosis occurs in people with advanced AIDS [13], when CD8 T cell deficiency allows a reactivation of latent T. gondii infection [14].
Serological tools are useful to monitor and characterise T. gondii infections. Specific anti-Toxoplasma IgM and/or IgG serum antibody levels (titres) provide an indication of the status of a T. gondii infection and its clinical relevance. Latent infection (sometimes referred to as 'latent toxoplasmosis'-although not always a disease state) [9] is usually associated with the presence of relatively low levels of specific IgG serum antibodies, whereas a subacute or acute infection linked to disease ('acute toxoplasmosis') [9,15] is associated with an initial increase in specific anti-Toxoplasma IgM serum antibody, followed by a four to 16-fold increase in specific IgG serum antibody over a period of 2-4 weeks [7,16,17]. Thus, serological tools underpin epidemiological investigations of the prevalence, status and/or dynamics of T. gondii infection and toxoplasmosis in human populations.
Estimating the prevalence and distribution of T. gondii infection using serodiagnostic tools and assessing associated risk factors for HIV + people will assist the development and implementation of measures/programs to prevent toxoplasmosis in this risk group [18]. Previous investigations to estimate the prevalence of T. gondii infections in HIV + people were limited to particular countries or geographical regions over short study periods (usually one year) [19][20][21][22][23]. Perhaps the most comprehensive investigation of HIV + people was a meta-analysis of studies published from 1987 to 2016 [24], which considered latent infection/toxoplasmosis, but not acute toxoplasmosis; furthermore, it did not assess risk factors linked to toxoplasmosis and its prevalence.
In the present study, we undertook a comprehensive, systematic review of observational studies to (i) estimate respective seroprevalence rates for latent and acute T. gondii infections in HIV + people at the global, regional and country levels; (ii) relate prevalence to socio-economic variables and CD4+ cell counts; and (iii) assess temporal changes in prevalence as well as risk factors for this group of people.

Materials and Methods
We followed the recommendations established by the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) [25]. All procedures, including the literature search, the critical appraisals of full articles/reports and of data quality as well as the analyses of data were performed by experienced researchers. Disagreements regarding study selection, data extraction and quality assessment were resolved in discussion with the first author (A.R.) to reach a consensus decision.

Search Strategy and Selection Criteria
A systematic review of the literature was performed using PubMed (through OVID), Scopus, Web of Science, SciELO and Embase to identify all peer-reviewed articles reporting data on the prevalence of latent and acute T. gondii infections in HIV + people, published between January 1980 and July 2020, without language or geographical restriction. We used relevant keywords for the searching of databases ( Figure S1). The search strategy was discussed with an expert medical librarian for optimal inclusion-sensitivity. To ensure that we identified 'grey' literature, such as papers in languages other than English, conference proceedings, and technical or governmental reports, we searched the Google Scholar database, and scrutinised reference lists of reviews and original articles.
The inclusion criteria were: (i) observational studies with original data used to estimate the seroprevalence of latent or acute T. gondii infection/toxoplasmosis in HIV + people; (ii) studies with a sample size of >30 individuals; and (iii) investigations to measure specific anti-Toxoplasma serum antibodies (IgM and/or IgG) or Toxoplasma antigens. For case-control studies, data were extracted only for HIV + individuals. Excluded were studies that (a) were not representative of a 'general population' of people with HIV infection (e.g., encephalitis cases, particularly those with toxoplasmic encephalitis, psychiatric patients and drug users); (b) lacked a detailed description of serodiagnostic method(s) used; (c) were comparative investigations of diagnostic methods; (d) had overlapping participation in multiple studies; and/or (e) were experimental studies, case-reports, case-series, reviews, systematic reviews or meta-analyses.
Here, HIV + people were assigned to two groups according to their specific anti-Toxoplasma serological profiles: Those who (i) were test-positive for specific anti-Toxoplasma IgG antibody with low avidity, (ii) were test-positive for both specific anti-Toxoplasma IgM and IgG antibodies, (iii) had a seroconversion with a four to 16-fold increase in specific IgG titre over a period of 2-4 weeks, or (iv) were test-positive for circulating Toxoplasma antigen in serum, were assigned to the "acute T. gondii infection" (AT-) group [7]. Those who were test-positive for specific anti-Toxoplasma IgG serum antibody in the absence of IgM serum antibody [26] were assigned to the "latent T. gondii infection" (LT-) group.

Data Extraction and Assessment of Quality
We extracted the following variables from each study: first author's name, country, city, year of publication, study design, number of participants, number of seropositive HIV + people in each of the AT-and LT-groups, and the type of diagnostic method(s) used. Countries were grouped according to the WHO-defined regions [27], income per capita and the human development index [HDI]. In order to identify associated risk factors for toxoplasmosis in HIV + people, we extracted data/information including: place of residence (urban/rural), gender (male or female), close contact with dogs or cats, contact with soil, and consumption of raw/undercooked meat, raw/unwashed vegetables and/or untreated drinking water. To evaluate the quality of the methodologies used in studies, we employed the Joanna Briggs Institute (JBI) prevalence critical appraisal tool [28].

Statistical Analysis
Analyses were performed using Stata statistical software (v.16 Stata Corp., College Station, TX, USA). For each individual study, prevalence was defined as the number of HIV + individuals in either the AT-or LT-group, divided by the total number of HIV + people studied. We used the variance of the study-specific prevalence estimates with the Freeman-Tukey double arcsine transformation prior to the pooling of data, employing the DerSimonian-Laird random-effects meta-analysis model to minimise the effect of studies with an extremely low or high reported prevalence [29][30][31]. The random-effects model [30] was selected in anticipation of significant variation in seroprevalence estimates among studies included here. Heterogeneity between studies was evaluated using Cochran's Q test and the I 2 index. I 2 values of 25%, 50% and 75% represented low, medium and substantial heterogeneities, respectively [32]. To explore factors that might influence heterogeneity, several subgroup and meta-regression analyses were performed, considering the following study features: (i) WHO region, (ii) country, (iii) type of diagnostic method employed, (iv) study design, (v) income level in a country, and (vi) HDI level in a country. Moreover, to calculate the global number of seropositive HIV + people in each of the AT-and LT-groups, we employed an established method [24] and used data from the WHO on the number of HIV + people worldwide and in different countries/regions in 2019 [33]. Subsequently, we multiplied this number by the calculated percentages of HIV + people in the AT-and LT-groups (95% CI) at the global and national levels. We did not undertake an assessment of publication bias, as this is not relevant for prevalence studies [34]. In all comparisons, a p-value of <0.05 was considered statistically significant.

Study Characteristics
Of 4024 published scientific reports or papers, 150 and 65 studies met the inclusion criteria for analyses of the seroprevalences in the AT-and LT-groups, respectively ( Figure S2). The 150 studies (155 data sets) surveyed 44,473 HIV + people from 49 countries, whereas the 65 studies (65 datasets) investigated 17,705 HIV + people from 32 countries. Relevant study characteristics are listed in Table S1. Most studies (n = 45) were from Africa, followed by the Western Pacific (n = 24) and the Eastern Mediterranean (n = 23). The lowest number of studies were from North America and the Caribbean (n = 12).  Figure 1A). Extrapolating to the global population, we estimated that 14,174,600 (12,658,600-15,690,600) of HIV + people had specific anti-T. gondii IgG serum antibody titres consistent with those expected of people with latent T. gondii infection (LT-group). These findings indicate clearly that countries in Africa have 11,873,400 (9,688,900-14,057,900) HIV + people with latent T. gondii infection, equating to~84% of HIV + people worldwide. Details pertaining to estimated seroprevalences in distinct WHO-regions and countries are given in Table 1 and Figure 1A.
Regarding risk factors, the findings showed that HIV + people who consumed raw/ undercooked meat (odds ratio [OR], 2.01; 95% CI, 1.19-3.9) and/or were in frequent contact with soil (OR, 3.01; 95% CI, 1.5-6.04) were more likely to be seropositive than those who consumed cooked meat and had limited contact with soil ( Figures S3 and S4 Figures S8 and S9). Other findings (Table 3 and Tables S10-S15) showed that females, people who lived in rural areas, consumed raw/unwashed vegetables or untreated water and/or were cat or dog owners were at a greater (but insignificant) risk of acquiring infection. No significant publication bias was seen, based on results achieved using the Egger's test (Table 3). In subgroup analyses, with respect to income level and HDI, the highest predicted seroprevalence for the LT-group was in low-income countries (58.2%, 46.2-69.8%) and the lowest in high-income countries (28.0%, 21.3-35.2%). The pooled prevalence in HIV + people in the LT-group in countries with low, medium, high and very high HDIs were 51.9% (42.8-60.8%), 29.3% (20.9-38.4%), 38.0% (22.6-33.1%) and 27.6% (22.6-33.1%), respectively ( Table 2). Meta-regression analyses revealed a significant decreasing trend in seroprevalence in countries with increasing levels of income per capita (C = −2.97 × 10 −6 ; p-value = 0.004) and/or increasing HDI (C = −0.473; p-value < 0.001) (Figure 2A,B). Studies published after 2005 indicated slightly higher seroprevalences than before, although the increasing

Discussion
This investigation provides seroprevalence estimates to infer burdens of latent or acute T. gondii infection in HIV + people at the global, regional and country levels, and sheds light on key socio-economic and immunological variables as well as risk factors for toxoplasmosis in this cohort of people.
The pooled seroprevalence rates estimated separately for latent (37.4;~14.2 million people) and acute (1.3%; 0.5 million people) T. gondii infections in HIV + people are similar to those reported (33.8% and 1.1%, respectively) for pregnant women worldwide [7,8]. The estimate for latent toxoplasmosis (37.4%) is slightly higher than the 35.8% reported previously for HIV + people by Wang et al. [23], who used data from a total of 74 studies of 25,989 HIV + people-i.e., 51% fewer studies and 42% fewer individuals than in the present investigation.
Here, the overall seroprevalence estimates for latent or acute T. gondii infection varied among regions, with the highest estimates in African and South-American countries and the lowest in the Western Pacific region, in agreement with previous global investigations of pregnant women [7,8]. The very high prevalence (up to~84%) of latent infection in countries or regions of Africa (particularly sub-Sahara), Latin America and the Caribbean compared with other parts of the world (Table 1) likely relates to socioeconomic factors (income and HDI status) and climatic factors (high environmental temperature and humidity); the public health and sanitary situation; the prevalence of stray cats and the extent of environmental contamination with T. gondii oocysts (i.e., soil and water); cultural or culinary habits, particularly the consumption of semi-cooked or raw meat; and the infectivity/virulence of particular T. gondii genotypes present [7,11,35,36].
The present study showed that the key risk factors for latent T. gondii infection in HIV + people are low CD4+ lymphocyte counts, age, frequent contact with soil, and the consumption of raw/undercooked meat, consistent with those proposed or documented in numerous previous publications and reviews [37][38][39][40][41]. Raw/undercooked meat is assumed to be a major source of T. gondii infection; however, as a risk factor, this depends on the kind of meat consumed and local culinary habits of meat preparation in a particular country or region; for example a report from China [42] indicated that the anti-T gondii seroprevalence in people who consumed beef or duck was higher than in than those who consumed pork, lamb and/or chicken. In European and American countries, lamb and pork are major dietary risk factors, whereas in the Middle East lamb is reported to be a key risk factor [37,40,43,44].
The increase in seroprevalence in HIV + people with age is simply explained by an increased probability of infection with, or exposure to, Toxoplasma over time [19,36]. We expect that one of the most important risk factors for toxoplasmosis in HIV + is a low CD4+ lymphocyte count. It is well documented that a decline in this count is a key reason for the reactivation of toxoplasmosis in HIV + individuals with latent T. gondii infection, ultimately leading to severe or fatal disease [13,14]. Clearly, this scenario applies also to other immunodeficiencies caused by transplantation, malignant cancers and/or chemotherapy [17,45,46], where accidental or subclinical infections with opportunistic pathogens can have devastating consequences.
While the findings of the present study are insightful, some results need to be interpreted cautiously. First, we found significant heterogeneity between studies, which is a common issue in epidemiological studies such as this one, and could not be adequately explained by the subgroup and meta-regression analyses. The sources of heterogeneity likely relate to study characteristics, including differences in study design, season, geographical distribution, sample size and/or detection methods, including differences in the quality and performance of these methods. Second, the vast majority of studies in the literature have used serological methods to detect latent or acute T. gondii infection; however, serum antibody responses to T. gondii in immunodeficient individuals with HIV infection might be low in some or many cases, indicating or suggesting that the prevalence rates calculated here might actually be an under representation of the actual prevalence status globally and regionally. While bioassays (in mice or cats) are considered as the gold standard for the definitive diagnosis of toxoplasmosis [47], these assays are usually not used for diagnosis in humans. Moreover, DNA-based methods do not commonly complement routine diagnosis in humans. Thus, serodiagnosis is the mainstay, but variation in the performance of serological assays (including diagnostic specificity and sensitivity; reproducibility and repeatability of results) will influence prevalence estimates. Third, relatively few studies met the selection criteria for inclusion; many studies lacked critical data and/or information or were clinically focused on individual toxoplasmosis cases. Thus, the prevalence rates estimated here should be considered as apparent, not actual. In addition, we acknowledge that, despite undertaking a comprehensive systematic search, published studies and data sets are not available for many parts of the world.

Conclusions
In spite of some limitations, this is the most comprehensive review and meta-analysis of available data sets to estimate the seroprevalence of T. gondii infections in HIV + people, and to identify associated risk factors. The inference that more than one third of all people with HIV infection likely have a latent T. gondii infection indicates that many of these individuals are at risk of developing 'reactivated' toxoplasmosis as a consequence of their immunodeficiency, and that a small percentage (~1%) are at risk of acquiring primary Toxoplasma infection. Current international guidelines [48] recommend that all HIV + individuals should be tested for specific anti-Toxoplasma IgG serum antibodies immediately after being diagnosed as having HIV infection, and that particularly individuals with no detectable anti-Toxoplasma serum antibodies should be counselled about the risks of them acquiring Toxoplasma and/or other opportunistic infections. These guidelines also emphasise that Toxoplasma-seropositive patients with CD4+ counts of <100 cells/µL should receive prescribed, routine treatment with trimethoprim + sulfamethoxazole or atovaquone to prevent toxoplasmic encephalitis and Pneumocystis jirovecii-associated pneumonia. However, it is noteworthy that a 2017 systematic review of toxoplasmosis chemoprophylaxis failed to identify any studies from Africa [49], while a recent study from Mozambique highlights the aspects of this illness that are comparable to a neglected tropical disease at the treatment, prevention and policy levels [50]. The high seroprevalence rates of T. gondii infection in lesser developed countries and regions of the world, including those in sub-Saharan Africa, call for action by governments and health authorities to introduce the routine serological monitoring, counselling, care and/or prophylactic treatment measures needed to prevent severe toxoplasmosis from developing in highly susceptible people living with HIV infection.

Supplementary Materials:
The following are available online at https://www.mdpi.com/article/10 .3390/microorganisms9102034/s1, Figure S1: Search strategy in databases, Figure S2: Flow diagram of the search strategy and study selection process, Figure S3: Forest plots of pooled odds ratio (OR) for latent T. gondii infection in HIV + people with respect to consumption of raw/under-cooked meat. HIV + people who did not consume raw/under-cooked meat were considered as the reference category to estimate OR, Figure S4: Forest plots of pooled odds ratio (OR) for latent T. gondii infection in HIV + people with respect to contact with soil. HIV + people without contact with soil were considered as the reference category to estimate OR, Figure S5: Forest plots of pooled odds ratio (OR) for latent T. gondii infection in HIV + people with respect to ages of 20-40 years compared to ages <20, Figure S6: Forest plots of pooled odds ratio (OR) for latent T. gondii infection in HIV + people with respect to the ages of 40-60 years compared to ages <20, Figure S7: Forest plots of pooled odds ratio (OR) for latent T. gondii infection in HIV + people with respect to ages >60 years compared to ages <20, Figure S8: Forest plots of pooled odds ratio (OR) for latent T. gondii infection in HIV + people with respect to age and number of CD4+ cells 200-500 compared to age and number of CD4+ cells >500, Figure S9: Forest plots of pooled odds ratio (OR) for latent T. gondii infection in HIV + people with respect to age and number of CD4+ <200 compared to age and number of CD4+ >500, Figure S10: Forest plots of pooled odds ratio (OR) for latent T. gondii infection in HIV + people with respect to gender (male and female). Female gender was considered as the reference category to estimate OR, Figure S11: Forest plots of pooled odds ratio (OR) for latent T. gondii infection in HIV + people with respect to type of residence (rural and urban). Living in an urban area was considered as the reference category to estimate OR, Figure S12: Forest plots of pooled odds ratio (OR) for latent T. gondii infection in HIV + people with respect to cat ownership. HIV + people who were not cat owners were considered the reference category to estimate OR, Figure S13: Forest plots of pooled odds ratio (OR) for latent T. gondii infection in HIV + people with respect to dog ownership. HIV + people who were not dog owners were considered the reference category to estimate OR, Figure  S14: Forest plots of pooled odds ratio (OR) for latent T. gondii infection in HIV + people with respect to consumption of raw/unwashed vegetables. HIV + people who did not consume raw/unwashed vegetables were considered as the reference category to estimate OR, Figure S15: Forest plots of pooled odds ratio (OR) for latent T. gondii infection in HIV + people with respect to drinking untreated water. HIV + people who drank treated water were considered as the reference category to estimate OR, Table S1: Main features of studies included on the seroprevalences of latent and acute Toxoplasma gondii infection (LT and AT, respectively) in HIV + people worldwide.