Toxoplasma gondii Infection in Immunocompromised Patients: A Systematic Review and Meta-Analysis

1The protozoan parasite Toxoplasma gondii can infect nearly all warm-blooded animals, including humans (Robert-Gangneux and Darde,; Liu et al., ). Approximately 30% of the world's population is estimated to be infected with T. gondii (Montoya and Liesenfeld, ). Humans become primarily infected by ingesting raw or undercooked meat containing viable tissue cysts, or by ingesting water or food contaminated with oocysts from infected cat feces (Baldursson and Karanis,; Meireles et al., ). In healthy humans, the infection with T. gondii is usually asymptomatic, but it can be fatal in the immunocompromised individuals, such as HIV/AIDS patients, cancer patients, and organ transplant recipients (Da Cunha et al.,; Pott and Castelo,; Agrawal et al.,; Lu et al., ).

2Toxoplasmosis of immunosuppressed individuals is most often the result of reactivation of latent infection, which presents neurological signs, including headache, disorientation, drowsiness, hemiparesis, reflex changes, and convulsions (Barratt et al.,; Robert-Gangneux and Darde, ). Acute acquired T. gondii infection in immunocompromised patients may also occur and involve multiple organs. Pneumonia, retinochoroiditis, and other disseminated systemic diseases, can also be seen, but are not as common as encephalitis in immunocompromised patients (Machala et al., ).

3An increased frequency of Toxoplasma encephalitis has been reported in AIDS patients, especially those with significant immunosuppression when CD4 T lymphocyte cell counts is <200 cells/μL, and T. gondii infection is regarded as an important opportunistic pathogen that lead to the death of AIDS patients (Luft et al.,; Jones et al., ). The cancer can also reactivate latent T. gondii infection during antitumor treatment process (Frenkel et al., ). A variety of malignancies, including lymphoma, leukemia, and myeloma, can reactivate toxoplasmosis (Maciel et al.,; Kojima et al., ). Transplantation of an organ from seropositive donor can activate latent infection in a seronegative recipient receiving immunotherapy (Chehrazi-Raffle et al., ). Transplantation of an organ from seronegative donor can also initiate fatal infection by activation of the latent infection in a seropositive recipient receiving immunosuppressive therapy. It seems that danger of transplanting an infected organ into a seronegative recipient is greater than that of transplanting a non-infected organ into a seropositive recipient (Chehrazi-Raffle et al., ). Fatal toxoplasmosis has been reported in heart, liver and bone marrow, haematopoietic stem cell transplant recipients (Castagnini et al.,; Caner et al.,; Stajner et al.,; Gajurel et al., ).

4Toxoplasmosis can be complicated and is considered a serious disease in immunocompromised patients, in which the reactivation of a latent infection can be fatal. The incidence of reactivated toxoplasmosis may rely on the prevalence and concentration of IgG antibodies (Robert-Gangneux and Darde, ). It is necessary to obtain information concerning the prevalence of T. gondii infection in different special populations worldwide. We conducted a global meta-analysis to assess the seroprevalence and odds ratios (ORs) of T. gondii infection in immunocompromised patients compared with those in control individuals.

5We reported this meta-analysis in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) statement (Moher et al., ). We searched PubMed, Embase, Google scholar, ScienceDirect, Chinese Web of Knowledge, Wanfang, and Chongqing VIP databases from inception to February 29, 2016, for all reports that possibly contained data for T. gondii prevalence in different immunocompromised populations. The databases were searched using the keywords “Toxoplasma gondii” and “toxoplasmosis” cross-referenced with “HIV,” “AIDS,” “acquired immune deficiency syndrome”, “cancer,” “tumor,” “malignancy,” “carcinoma,” “transplantation,” “organ grafting,” “immunodeficiency,” and “immune deficiency.” We included studies without language limitation.

6We systematically searched the scientific literatures for case-control, cohort, and cross-sectional studies that reported T. gondii infection in immunocompromised individuals, stratified by one of the following criteria: population with HIV/AIDS or without HIV/AIDS; population with cancer or without cancer; transplant or non-transplant population. Studies were excluded if they were reviews, repeated studies, or animal studies. Studies were excluded if they provided the final result without raw data. Studies were excluded if the sample size from one of the two groups was <30.

7All identified titles and abstracts were carefully examined by two independent reviewers (HHL and HYM). The full text of articles considered as potentially relevant based on title and abstract were independently examined by the same two reviewers. Any disagreements with the selected studies were resolved by discussion and the involvement of another two authors (ZDW and QL).

8The following information was extracted from each study: first author, publication year, country of the study, the number of patients and control, diagnostic methods, and demographic characteristics. Two reviewers (ZDW and YZL) independently extracted the data and reached a consensus after a discussion on the controversial literatures.

9The quality of the included publications was assessed based on the criteria (Liu et al.,; Speich et al., ). These criteria were created based on the Grading of Recommendations Assessment, Development and Evaluation method (Atkins et al., ), and including the diagnostic approach of T. gondii infection and matching of case and control subjects (Table 1). A scoring approach was used for grading, and up to 11 points assigned to each study. Studies that were awarded 6–11 points were considered to be of high quality, 4–5 points were moderate quality, whereas lower scores indicated low quality.

10We estimated prevalence of T. gondii infection by pooling of data from each study. Data were pooled with a DerSimonian-Laird random-effects model (DerSimonian and Laird,; Borenstein et al., ), whose difference was compared using Wilcoxon two-sample test or t-test. The risk of T. gondii infection in patient and control groups was estimated by odds ratio (OR). It was considered statistically significant when p < 0.05. In the forest plots, OR > 1 showed a risk effect and OR < 1 showed a protective effect. Statistical heterogeneity of results was appraised using a x2-based Q-test and I2 statistic. The heterogeneity was considered not significant only when p > 0.1 and I2 < 50%. The fixed-effects model was used when literature heterogeneity not existed; otherwise, the random-effects model was employed. Sensitivity analysis was performed by modification of the inclusion criteria of this meta-analysis. The analysis was conducted using Stata software version 12.0 (Stata Corporation, College Station, TX, USA). The publication bias was considered significant when p-value of Begg's test and Egger's test was <0.05.

11As shown in Figure 1, the literature search yielded 11,799 relevant studies, which included 2,434 duplicates. After a careful examination of each article's title and abstract, 493 were considered as having potential value, and the full texts were retrieved for detailed evaluation. A total of 422 potentially relevant articles were excluded from this meta-analysis after consulting the full text. Of these, 273 articles did not present sufficient data that required, or not conform with the included criteria; 135 had prevalence without raw data; 9 had unmatched control populations; the sample size in three articles was <30; there were two publications whose full texts were not retrieved. One additional publication regarding T. gondii infection in HIV patients was identified through the second search on Feb, 29, 2016. Finally, a total of 72 publications were included for our meta-analyses.

12Characteristics of the included publications are listed in Tables 2–4. In brief, 38 publications described T. gondii infection in HIV/AIDS patients, 28 articles investigated T. gondii infection in cancer patients, whereas 6 studies reported T. gondii infection in transplant patients. The identified studies were conducted worldwide (Figure 2). In terms of epidemiological design, 51 of the included publications were case-control studies, 17 were cross-sectional studies, and four were cohort studies. Thirty-nine papers were written in English, 29 were in Chinese, two in French (Maiga et al.,; Gamba et al., ), and one each in Spanish (Gongora-Biachi et al., ) and in Croatian (Dakovic-Rode et al., ). There were 6 papers whose raw data were extracted from the abstract (Ryan et al.,; Gongora-Biachi et al.,; Sukthana et al.,; Uneke et al.,; Akanmu et al.,; Manouchehri Naeini et al., ). The oldest study was conducted in 1987 (Quinn et al., ). Totally, 40 datasets investigated T. gondii infection in HIV/AIDS patients, and 29 datasets examined T. gondii infection in cancer patients, whereas only six datasets studied T. gondii infection in transplant recipients.

13According to our criteria, eight publications were of high quality (>6 points), 43 publications had quality scores of 4–6 points indicating moderate quality, whereas the remaining 21 publications were of low quality (<4 points).

14The estimated pooled prevalence of T. gondii infection in the HIV/AIDS patients and control population was 42.1% (95% CI, 34.0–50.2%) and 32.0% (95% CI, 24.0–40.1%), respectively (p < 0.05); the prevalence in cancer patients and control was 26.0% (95% CI, 20.5–31.5%) and 12.1% (95% CI, 9.5–14.8%), respectively (p < 0.001); and the prevalence in transplant recipients and its control was 42.1% (95% CI, 27.1–57.2%) and 34.5% (95% CI, 17.1–51.9%), respectively (p = 0.59). The results are shown in Supplementary Figures 1–6.

15Forest plots on the association of immunosuppressed populations with T. gondii infection are presented in Figures 3–5. The estimated pooled random effects ORs of HIV/AIDS, cancer, and transplant patients compared with their controls were 1.92 (95% CI, 1.44–2.55), 2.89 (95% CI, 2.36–3.55), and 1.51 (95% CI, 1.16–1.95) for infection with T. gondii. However, the heterogeneity analysis showed that there was a relatively high-level heterogeneity in our meta-analysis of HIV/AID patients (Q = 401.6, I2 = 90.3%, p = 0.000) and cancer individuals (Q = 76.4, I2 = 63.4%, p = 0.000), and no heterogeneity was found in transplant recipients (Q = 8.0, I2 = 37.3%, p = 0.157).

16We also analyzed all the data, showing that the estimated pooled prevalence of T. gondii infection in immunocompromised patients and the controls was 35.9% (95% CI, 31.0–40.8%) and 24.7% (95% CI, 20.5–28.8%, p < 0.001), with an OR of 2.24 (95% CI, 1.87–2.69).

17All subgroup analysis, including those of geographical distribution, country income, published years, sample size, detection methods, study design, and population, did not show any significant differences between the respective groups, as indicated by overlapping 95% CIs, with the exception of subgroups based on geographical distribution and income level (Supplementary Table 1). For example, the odd of T. gondii infection in HIV/AIDS patients in Asia (OR = 2.77, 95% CI, 1.58–4.87) was significantly higher as comparison with that in Latin America (OR = 1.19, 95% CI, 0.90–1.56) and in Europe (OR = 0.97, 95% CI, 0.65–1.46); the odd of T. gondii infection in cancer patients in Asia (OR = 3.07, 95% CI, 2.51–3.76) was significantly higher, compared with that in Oceania (OR = 1.42, 95% CI, 0.80–2.54). Additionally, higher odds of T. gondii infection in both HIV/AIDS and cancer patients were found in middle-income and low-income countries, compared with that of high-income countries.

18Begg and Egger tests were used to evaluate the publication bias. No significant bias was revealed in HIV/AIDS- or transplant-related publications, but significant bias was observed in cancer-associated publications (p < 0.05, Supplementary Table 1, Supplementary Figure 7).

19A sensitivity analysis was conducted by excluding one single study each time to find out whether modification of the inclusion criteria of this meta-analysis had an effect on the final results. All the results were not materially altered (data not shown).

20Our meta-analysis focused on T. gondii IgG antibodies, which are a marker of lifetime exposure to toxoplasmosis, whereas IgM antibodies are a marker of acute or recent infection, or also potentially persistent infection or reinfection with a different genotype (Sharma et al.,; Dzitko et al., ). During extraction of data in this study, the IgM antibodies against T. gondii in immunocompromised patients were also collected (Supplementary Tables 2–4). Due to insufficient data on HIV/AIDS and transplant patients, we only analyzed T. gondii IgM in cancer patients and the control, showing a prevalence of 11.4% (95% CI, 8.1–14.7%) in cancer patients and 2.7% (95% CI, 1.5–4.0%, p < 0.01) in its control group and OR of 2.65 (95% CI, 2.04–3.45, Supplementary Figures 8–10). The results also confirmed that the immunocompromised patients were associated with significantly higher odds of recently acquired T. gondii infection.

21T. gondii has been suggested as an important opportunistic pathogen in immunocompromised patients (Walker and Zunt, ). The infection in healthy (immunocompetent) people is usually self-limited and asymptomatic, resulting in chronic infection of tissue cysts that can lie dormant, probably for the entire lifetime of the hosts. However, immunocompromised individuals, such as HIV/ADIS patients, cancer patients with chemotherapy, and transplant recipients, are at risk of developing Toxoplasma encephalitis, myocarditis, or pneumonitis, due to reactivation of the chronic infection. For example, approximately 30–40% of HIV co-infected immunocompromised individuals with T. gondii develop encephalitis (Walker and Zunt, ).

22The associations of HIV and seroprevalence of T. gondii infection are varied in the world (Grant et al., ). Some reports showed higher prevalence of T. gondii infection in HIV-infected patients compared to non-infected individuals, whereas others did not find any differences between the two groups (Sukthana et al.,; Galvan-Ramirez Mde et al., ). This global systematic review and meta-analyses were conducted to quantify the prevalence and ORs of T. gondii infection in immunocompromised individuals compared with those in control individuals.

23Subgroup analyses comparing published year, sample size, detection method, study design, country income, and population revealed non-significant differences, but high odds were found for T. gondii infection in HIV/AIDS patients in Asia and Africa as comparison with that of America and Europe, and in cancer patients in Asia compared to that in Oceania (Supplementary Table 1). However, only one or two studies examined the association of immunocompromised patients with T. gondii infection in these regions. Thus, no meta-analysis could be done and no firm conclusion should be drawn. Most studies were conducted in the countries of Asia. Our analyses further demonstrated that the studies are geographically clustered, with few studies in Latin America, Europe, and Oceania (Figure 2).

24The presence of heterogeneity was observed in HIV/AIDS and cancer patients, but subgroup analyses did not explain the specific causes of heterogeneity, which may come from various sources, including geographical distribution, published years, sample size, detection methods, study design, or populations. Without meta-regression or additional subgroup analysis that requires a large number of studies, it is difficult to investigate the causes of heterogeneity. The presence of heterogeneity shows that pooled results are averaging multiple related, but different effects (Strunz et al., ).

25In fact, higher prevalence of T. gondii infection in HIV/AIDS patients has been reported in many countries, such as Nigeria, Mali, Ethopia, India, China, and Thailand (Maiga et al.,; Wanachiwanawin et al.,; Akanmu et al.,; Daryani et al.,; Uppal et al.,; Shen et al., ). The present study provided robust evidence that support the conclusion, and demonstrated that HIV/AIDS patients are associated risk factors (OR = 1.92, 95% CI 1.44–2.55) for T. gondii infection. The data were derived from 38 publications from 20 countries, which included 10,028 HIV/AIDS patients and 12,334 control people (Table 2).

26A recent study reported T. gondii infection in Chinese cancer population, with a prevalence of 20.6% in cancer patients and 6.3% in the control (OR = 3.9) (Jiang et al., ). Our meta-analysis also included the data of other countries, such as Australia, Iran, and Turkey (Table 3), which involved 7,011 cancer patients and 9,254 control people, therefore, the results would be more reliable. However, of the included 28 publications, 21 were written in Chinese, resulting in significant publication bias.

27There are many case reports of toxoplasmosis in transplant recipients, including haematopoietic cell (Barcan et al., ), heart (Gajurel et al., ), liver (Hamza et al., ), and kidney (Petty et al., ) transplant patients. A previous study reported a higher prevalence of T. gondii infection in renal transplant recipients (Soltani et al., ). In the present study, we analyzed T. gondii infection in 671 transplant patients (heart, kidney, and liver) and 628 control people from five countries (Table 4), revealing no significant difference of T. gondii infection between the two groups, but showing that transplant population is an risk factor (OR = 1.51, 95% CI, 1.16–1.95) for T. gondii infection. Interestingly, it was found that 14.3% of renal transplant recipients were detected positive for T. gondii infection in the first year of transplantation, and the prevalence increased to 85.7% in 1 year post-transplantation (Aufy et al., ).

28There are several limitations in this meta-analysis, which may affect the results. First, a number of potentially relevant studies were identified through our systematic review, but not all the underlying data were available. Therefore, though most of these studies might not have relevant data, there is a certain risk to miss some eligible data.

29Second, based on our scoring system, most studies were of moderate or even relatively low quality. This finding is mainly due to the epidemiological design of the studies; most were cross-sectional in nature. The differences between the study groups also included ages, lifestyles, and geographical conditions, which all contribute to the difference of T. gondii infection between the patient and control groups (Minbaeva et al.,; Walle et al.,; Wang et al., ). The cluster randomized controlled trial would provide high-quality data, but their implementation is more difficult. An alternative way to generate high-quality data would be using a time-series approach as a study design (Speich et al., ).

30Third, diagnosis of T. gondii infection in immunocompromised patients is difficult. Though detection of the parasite by microscopy and bioassays is considered as the gold standard for diagnosis of toxoplasmosis, its clinical diagnosis more likely relies on serological methods (Liu et al., ). However, the serological methods may be unreliable in the immunocompromised individuals, whose immune system has been impaired, and cannot produce enough antibodies (Lewis et al., ). In the identified studies, all the detection methods were serological, including indirect hemagglutination (IHA), dye test (DT), immunochromatographic test (ICT), and enzyme-linked immunosorbent assay (ELISA) (Tables 2–4). Thus, by the reason of lack of specific antibody, the detected results would be lower than the actual prevalence in immunocompromised patients, including HIV/AIDS patients, cancer patients, and transplant recipients (Saadatnia and Golkar, ).

31Fourth, insufficient data about further relevant factors on T. gondii infection (e.g., age, cancer type, transplanted organ) were available for subgroup analysis.

32In summary, our global meta-analysis shows a higher prevalence of T. gondii infection in immunocompromised patients, and demonstrates that the immunocompromised individuals, including HIV/AIDS patients, cancer patients, and transplant recipients, were associated with higher odds of T. gondii infection. Therefore, a routine serological screening test for T. gondii infection is suggested to be conducted in immunocompromised patients in endemic area, or patients with no proper chemoprophylaxis and/or HAART. Patients with a positive result are at risk of reactivation of the infection, while patients with a negative result should be informed to prevent primary infection. Health education, particularly toward avoiding eating raw and undercooked meat, and avoiding contact with cats' feces should also be considered.

33QL was responsible for the idea and concept of the paper. ZW and QL analyzed the results. HL, ZM, HM, ZL, FW, ZY, and BX collected and analyzed the data. QL and XZ wrote the manuscript. All authors contributed to the manuscript editing and approved the final manuscript.

34This work was supported by the National Natural Science Foundation of China (31672542, 31472183, 31372430 and 31230073) and the Special Fund for Agro-scientific Research in the Public Interest in China (201303042).

35The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.