The prevalence of cryptococcal antigen (CrAg) and benefits of pre-emptive antifungal treatment among HIV-infected persons with CD4+ T-cell counts < 200 cells/μL: evidence based on a meta-analysis

Background Current WHO guidelines (2018) recommend screening for cryptococcal antigen (CrAg) in HIV-infected persons with CD4+ T cell counts< 100 cells/μL, followed by pre-emptive antifungal therapy among CrAg positive (CrAg+) persons, to prevent cryptococcal meningitis related deaths. This strategy may also be considered for those persons with a CD4+ T cell count of < 200 cells/uL according the WHO guidelines. However, there is sparse evidence in the literature supporting CrAg screening and pre-emptive antifungal therapy in those HIV-infected persons with this CD4+ T cell counts< 200 cells/μL. Method We conducted a meta-analysis using data extracted from randomized controlled studies (RCTs) and cohort studies found in a search of Pubmed, Web of Science, the Cochrane Library and the EMBASE/MEDLINE database. Results The pooled prevalence of CrAg positivity in HIV-infected persons with CD4+ T cell counts< 200 cells/μL was 5% (95%CI: 2–7). The incidence of CM in CrAg+ persons was 3% (95%CI: 1–6). Among those CrAg+ persons who did not receive pre-emptive treatment, or those who received placebo, the incidence of CM was 5% (95%CI: 2–9), whereas the incidence of CM among those who received pre-emptive antifungal therapy was 3% (95%CI: 1–6), which is a statistically significant reduction in incidence of 40% (RR: 7.64, 95%CI: 2.96–19.73, p < 0.00001). As for persons with CD4+ T cell counts between 101 ~ 200 cells/μL, the risk ratio for the incidence of CM among those receiving placebo or no intervention was 1.15, compared to those receiving antifungal treatment (95%CI: 0.16–8.13). Conclusions In our meta-analysis the incidence of CM was significantly reduced by pre-emptive antifungal therapy in CrAg+ HIV-infected persons with CD4 <  200 cells/μL. However, more specific observational data in persons with CD4+ T cell counts between 101 ~ 200 cells/μL are required in order to emphasize specific benefit of CrAg screening and pre-emptive antifungal treating in CrAg+ persons with CD4+ T cell counts < 200 cells/μL.


Background
Cryptococcal meningitis (CM) continues to cause significant mortality in HIV-infected individuals [1,2], and results in 181,100 deaths globally each year [3]. In resource-limited regions such as sub-Saharan Africa, 15% of HIV-related deaths are due to CM [3]. However, it is possible to detect cryptococcal antigen (CrAg) in blood several weeks to months (22 days on average) before the onset of signs and symptoms of meningitis [4,5], and thus, the presence of CrAg in blood may be used as a marker for initiation of pre-emptive antifungal therapy in HIV-infected individuals with low CD4+ T cell counts. Previous studies have emphasized that preemptive antifungal therapy in CrAg+ persons is imperative to prevent death [6][7][8]. The prevalence of CrAg positivity among HIV-infected individuals can be considerable, ranging between 1 to 16% in several African and Southeast Asian countries [9], and among persons with CD4+ T cells counts< 100 cells/μL, the prevalence of CrAg positivity averages 7%, with regional variations in prevalence [3]. CrAg positivity resulted in a 20% increase in mortality after antiretroviral therapy (ART) initiation [10] if fluconazole therapy was not initiated prior to ART initiation, and the risk of CM in CrAg+ persons may be as high as 25% during the first year of ART, when fluconazole pre-emptive therapy is not prescribed for these patients [11,12].
According to the 2018 version of the WHO guidelines, routine CrAg screening and pre-emptive antifungal therapy are recommended in treatment-naive HIV persons with CD4+ T cell counts< 100 cells/μL [13]. The guidelines also state that these strategies may also be considered for HIV-infected persons with CD4+ T cell counts< 200 cells/μL [13]. We therefore conducted a meta-analysis to investigate the prevalence of CrAg positivity in HIV-infected patients, and the benefit of preemptive antifungal treatment in HIV-infected persons with CD4+ T cell counts< 200 cells/μL.

Search strategy and article screening
We searched relevant English articles in Pubmed, Cochrane Library, MEDLINE/EMBASE and Web of Science from inception until the end of March 20th 2020. The search terms we used were as follows: "acquired immunodeficiency syndrome", "HIV", "AIDS", "cryptococcosis", and "prophylaxis". We combined these terms by using "and" or "or". To avoid missing significant articles, we also screened references of previous meta-analyses and their included studies for eligibility.
Two reviewers (Y L, Y Q) independently screened all obtained articles by titles and abstracts. After removing ineligible articles by referring to our inclusion and exclusion criteria, the remaining articles were further selected for full-text reviewing. Data extraction and quality assessment

Inclusion and exclusion criteria
The data we extracted included first author, publication year, type of study, study duration, study location, total number of study subjects, baseline CD4+ T-cell counts, age, CrAg screening methods, diagnostic methods for CM, CM events, death events, adverse drug effects, and other opportunistic infections. The JBI (Joanna Briggs Institute) Critical Appraisal Checklist for Cohort Studies was used as a quality assessment tool for cohort studies [14]. The potential bias risk of RCTs was assessed using the Cochrane "risk of bias" tool [15].

Data analysis
Statistical analysis of data related to proportion of CrAg positivity, the incidence of CM, and all-cause mortality were performed by STATA 14 (Statacorp, Texas, USA) with a 95% confidence interval (95%CI). We used random-effects or fixed-effects models in Review manager 5.3 (The Nordic Cochrane Center, Copenhagen) to compare the incidence of CM and all-cause mortality in CrAg+ persons. We evaluated statistical heterogeneity through visual inspection of forest plots. Statistical heterogeneity was also assessed using I 2 statistics [16], which was considered non-negligible if I 2 > 50%. Herein, random-model was applied if I 2 > 50% and fixed-model was used when I 2 < 50% [17]. Reporting bias was assessed by examining the asymmetry of funnel plots [16].
The study was registered at the International Prospective Register of Systematic Reviews (PROSPERO), and the registration number is CRD42018110980.

Results
In total, 517 articles were obtained from 4 databases, among which 295 were from Pubmed, 111 were from Web of Science, 13 were from Cochrane Library, and 98 were from MEDLINE/EMBASE. Eighty-four of the 517 articles were RCTs or cohort studies. Additional 12 articles (RCTs or cohort studies) were extracted from references of previous meta-analyses and their included studies, as shown in Fig. 1.
All the 96 RCTs or cohort studies were included for screening. Initially, nine articles (six from Web of Science and three from MEDLINE/EMBASE) were found to be duplicated, and were therefore excluded from the 96 articles. After screening titles and abstracts, 43 of the remaining 87 articles were excluded. Subsequently, 35 articles were excluded from the remaining 44 articles after screening the full-text of each study, among which ten articles only included patients with CD4+ T-cell counts≤100 cells/μL, one article was a news report, three articles reported patients with cryptococcal disease, two articles reported HIV-negative patients with cryptococcal antigenemia, four articles reported data from patients with CM or asymptomatic CM, six articles reported data from HIV-infected patients with negative CrAg, one article reported on the epidemiology of cryptococcosis, and eight articles did not report CD4+ T cell counts or primary outcomes. Finally, a total of 9 articles were included in our meta-analysis.
The characteristics of the 9 included studies were shown in Table 1. Our assessment of quality and potential risk bias in these studies indicated that the following factors could contribute to clinical and methodological heterogeneity, including: (1) the confounding factors or subject recruiting or incomplete follow-up in one of the 8 cohort studies, (2) the unclear risk of attrition in the RCT, and (3) the unclear risk of reporting and other bias in the RCT, as shown in Supplementary Table 1 and  Supplementary Table 2 Fig. 2.
Six studies reported the incidence of CM among persons who received antifungal therapy (922 persons with CD4+ T cell counts< 200 cells/μL in four studies; 166 persons with CD4+ T cell counts< 150 cells/μL in two studies) and four studies reported the incidence of CM among persons who received placebo or no intervention (946 persons with CD4+ T cell counts< 200 cells/μL in three studies; 146 persons with CD4+ T cell counts< 150 cells/μL in one studies). The incidence of CM of 1088 persons receiving antifungal therapy was 3% (95%CI: 1-6; P = 0.037; I 2 = 57.7%), whereas the incidence of CM of 1092 persons in nine studies who received placebo or no intervention was 5%, which equates to a 40% reduction in CM incidence in persons receiving antifungal therapy (95%CI: 2-9; P = 0.015; I 2 = 71.5%), as shown in Table 2 and Supplementary Figure 1c and d.
Four studies compared the incidence of CM between 1030 persons receiving azoles and 1050 persons receiving placebo or no intervention (1785 persons with CD4+ T cell counts< 200 cells/μL in three studies; 295 persons with CD4+ T cell counts< 150 cells/μL in one study). We found that the risk ratio of CM events among persons who received placebo or no intervention was 7.64 times higher than that of those who received antifungal therapy (95%CI: 2.96-19.73; P < 0.00001; I 2 = 0%), as shown in Fig. 3.
In addition, we estimated and compared the prevalence of CrAg positivity, the incidence of CM and all-  Table 3.

Discussion
Several meta-analyses have been conducted in the past designed to evaluate the necessity of CrAg screening and administration of pre-emptive antifungal treatment among HIV-infected CrAg+ persons with varying low CD4 levels. For example, Temfack et al investigated the effectiveness of CrAg detection and the initiation of pre-emptive fluconazole treatment in HIV-infected persons with cryptococcal antigenemia and CD4+ T cell levels< 100 cells/μL [16]. Their results suggested that administration of fluconazole pre-emptive therapy to CrAg+ persons greatly reduced the risk of incident CM, and may have specific survival benefits [16]. Another metaanalysis conducted by Ssekitoleko et al also suggested that in resource-limited settings, CrAg+ persons should routinely receive primary antifungal prophylaxis [25], but they failed to clarify at which specific CD4+ T cell count antifungal prophylaxis should be initiated. Ford et al's [26] meta-analysis only reported the combined prevalence of cryptococcal antigenemia among HIVinfected persons with CD4+ T cell counts≤100 cells/μL, and with CD4+ T cell counts between 101~200 cells/ μL. Importantly, their study did not mention whether pre-emptive antifungal treatment was necessary or effective among HIV-infected persons with cryptococcal antigenemia at these two CD4+ T cell count strata. From the above studies, it may be gathered that the prudence and benefits of CrAg screening and pre-emptive antifungal therapy remain unclear at higher CD4+ T cell counts. The objective of our meta-analysis was to investigate the prevalence of cryptococcal antigenemia in HIVinfected patients with CD4+ T cell counts< 200 cells/μL, and the potential benefit of pre-emptive antifungal therapy among HIV-infected persons with cryptococcal antigenemia and CD4+ T cell counts< 200 cells/μL.
The pooled prevalence of CrAg positivity in HIVinfected persons with CD4+ T cell counts< 200cells/μL was 5% (5 studies) in our meta-analysis, which was similar to 6% (31 studies) among HIV-infected persons with CD4+ T cell counts< 100cells/μL in Temfack's metaanalysis [16] and 6.5% (60 studies) among HIV-infected persons with CD4+ T cell counts < 100 cells/μL in Ford's meta-analysis [26]. Therefore, antifungal Our results have demonstrated that, in persons with CD4+ T cell count< 200 cells/μL, the risk ratio of CM events among those who received placebo or no intervention was significant higher than those who received antifungal therapy, suggesting that antifungal prophylaxis significantly reduce the risk of CM events in CrAg+ persons with a higher CD4+ T-cell counts. However, the very limited data among persons with CD4+ T cell counts between 101~200 cells/μL restricted our capacity to investigate it further. Thus, more specific data are needed to demonstrate the benefit of antifungal treatment in HIV-infected persons with CD4+ T cell counts between 100 and 200 cells/μL, and warrants further investigation.
No significant difference in all-cause mortality was found in our meta-analysis among CrAg+ persons who received pre-emptive antifungal therapy versus placebo or no intervention. This is a somewhat surprising outcome, and the reason of this may be associated with the discrepant sample sizes in these two groups (396 vs. 1092).
We considered the following possible reasons for clinical and methodological heterogeneity: discrepancies in follow-up time for reporting CM events and death events, variations in drug dosing, regimens, or drug class of prescribed antifungal therapy, ART status of subjects, and risk of bias. For example, the study durations ranged from104 weeks to 6 years, and the dosing of azole antifungal treatments ranged from 100 mg/d to 900 mg/d. With regards to reporting bias, it is possible that the unformed funnel plot for all-cause mortality could be a consequence of the varied ART status of study participants, different dosage regimens and duration of treatment and the different follow-up periods in each of the individual studies.
There are some limitations in our study. Firstly, the data supporting the association between prevalence of CrAg positivity and occurrence of adverse outcomes in HIV-infected persons with CD4+ T-cell counts between 100 and 200 cells cells/μL is sparse. Secondly, there exists a paucity of new data regarding CrAg positivity prevalence, CM incidence, and all-cause mortality in HIV-infected persons with CD4+ T-cell counts< 200 cells/μL since 2015 [27], and our pooled outcome analyses relied heavily on older studies, which may be less applicable to the modern test-and-treat era. And thirdly, the dosage and durations of azole therapy was not assessed in our meta-analysis. The preceding limitations may contribute to the clinical and methodological heterogeneity in our study.