Increased interferon-γ levels and risk of severe malaria: a meta-analysis

Interferon (IFN)-γ contributes to the pathogenesis of severe malaria; however, its mechanism remains unclear. Herein, differences in IFN-γ levels between patients with severe and uncomplicated malaria were evaluated using qualitative and quantitative (meta-analysis) approaches. The systematic review protocol was registered at PROSPERO (ID: CRD42022315213). The searches for relevant studies were performed in five databases, including PubMed, Scopus, Embase, MEDLINE and Web of Science, between 1 January and 10 July 2022. A meta-analysis was conducted to pool the mean difference (MD) of IFN-γ levels between patients with severe malaria and those with uncomplicated malaria using a random-effects model (DerSimonian and Laird method). Overall, qualitative synthesis indicated that most studies (14, 58.3%) reported no statistically significant difference in IFN-γ levels between patients with severe malaria and those with uncomplicated malaria. Meanwhile, remaining studies (9, 37.5%) reported that IFN-γ levels were significantly higher in patients with severe malaria than those in patients with uncomplicated malaria. Only one study (4.17%) reported that IFN-γ levels were significantly lower in patients with severe malaria than those in patients with uncomplicated malaria. The meta-analysis results indicated that patients with severe malaria had higher mean IFN-γ levels than those with uncomplicated malaria (p < 0.001, MD: 13.63 pg/mL, 95% confidence interval: 6.98–20.29 pg/mL, I2: 99.02%, 14 studies/15 study sites, 652 severe cases/1096 uncomplicated cases). In summary, patients with severe malaria exhibited higher IFN-γ levels than those with uncomplicated malaria, although the heterogeneity of the outcomes is yet to be elucidated. To confirm whether alteration in IFN-γ levels of patients with malaria may indicate disease severity and/or poor prognosis, further studies are warranted.

Data extraction. Two authors (KUK and AM) carried out the data extraction, and data from the included studies were cross-checked by another author (MK). As a result, the following data were extracted to the spreadsheet: the name of the first author, publication year, study location, period of data collection, number of patients, age group, percentage of male participants, data on IFN-γ levels (qualitative data, mean with standard deviation or median with range for meta-analysis), parasite density, the technique used for detecting malaria parasites and technique used for measuring IFN-γ. Quality of the included studies. The methodological quality of the included studies was evaluated using the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) checklist for cross-sectional, prospective observational, and case-control studies 29 .
Data syntheses. Data syntheses included qualitative and quantitative syntheses. A qualitative synthesis was the narrative synthesis of the difference in IFN-γ levels between patients with severe and uncomplicated malaria. A quantitative synthesis (meta-analysis) was conducted to pool the mean difference (MD) of IFN-γ levels between patients with severe and those with uncomplicated malaria. The random-effects model was employed to pool the MDs using the DerSimonian and Laird method 30 . The mean and standard deviation were estimated from the median and range as described previously 31 . Comparing the mean and standard deviation between groups of participants was carried out using a protocol described previously 32 . If the standard deviation was unavailable, a value was borrowed from other studies (with the lowest standard deviation) in the same meta-analysis 33 . The heterogeneity of the estimated effects between studies was assessed using Cochrane Q and I 2 tests. A Cochrane Q test result with a significant value (p) less than 0.1 or an I 2 result of more than 50% indicated heterogeneity in the estimates of effect between studies. When heterogeneity was revealed, a meta-regression analysis was performed to identify the source(s) of heterogeneity in the effect estimates. Then, a subgroup analysis was performed to determine the differences in the effect estimates between the subgroups of interest. Finally, a sensitivity analysis using the leave-one-out method 34 was performed to test whether the meta-analysis was robust. Sensitivity analysis between studies that reported mean/standard deviation and median/range and studies that reported mean without standard deviation was conducted to assess the effect of changing the assumptions made. The sensitivity analysis between studies that reported mean/standard deviation and those with median/range of IFN-γ levels in patients with severe and uncomplicated malaria was conducted using a subgroup analysis. Publication bias was assessed by visually assessing funnel plot asymmetry and Egger's test for small-study effects. If publication bias Quality of the included studies. The quality of the included studies was determined using the STROBE checklist. The assessment results showed that seven of nine prospective observational studies (77.8%) were of high quality, while three were of moderate quality (22.2%). Eight of the ten case-control studies (70%) were of high quality, while two were of moderate quality (30%). Two prospective cohort studies were of high quality, and one cross-sectional study was of moderate quality (Table S3).
Overall, the results demonstrated that patients with severe malaria had higher mean IFN-γ levels than those with uncomplicated malaria (p < 0.001, MD: 13.63 pg/mL, 95% CI 6.98-20.29 pg/mL, I 2 : 99.02%, 14 studies/15 study sites, 652 severe cases/1096 uncomplicated cases, Fig. 2). Because of the high heterogeneity of the effect estimates among the included studies, a meta-regression analysis incorporating study design, location (continent), age group and technique used to measure the IFN-γ levels was conducted to identify whether these covariates were the source(s) of heterogeneity. The findings showed that study design (p < 0.001), age group (p = 0.001), location (continent, p < 0.001) and technique used to measure the IFN-γ levels (p < 0.009) were sources of heterogeneity in the effect estimates among the included studies. Therefore, subgroup analyses of study design, age group, location (continent) and technique used to measure the IFN-γ levels were performed.
The subgroup analysis by the study design showed that higher mean IFN-γ levels were found in patients with severe malaria than in those with uncomplicated malaria in prospective observational studies (MD: 25 Fig. 3).
The subgroup analysis by the age of the enrolled patients showed no differences in mean IFN-γ levels in patients with severe malaria and those with uncomplicated malaria in children (MD 5.76 pg/mL, 95% CI Table 2. Differences in IFN-γ levels between patients with severe and uncomplicated malaria based on qualitative data. *Results based on the statistical tests by included studies.
The subgroup analysis by the technique for IFN-γ measurement showed higher mean IFN-γ levels in patients with severe malaria than in those with uncomplicated malaria among studies using ELISA for IFN-γ measurement (MD: 26.79 pg/mL, 95% CI 15.26-38.31 pg/mL, I 2 : 99.05%, ten studies with 11 study sites). However, no differences in mean IFN-γ levels between patients with severe and those with uncomplicated malaria were found among studies using bead-based assays for IFN-γ measurement (MD: 0.95 pg/mL, 95% CI − 12.29-14.18 pg/ mL, I 2 : 99.10%, four studies, 256 severe cases/603 uncomplicated cases, 396 severe cases/492 uncomplicated cases, Fig. 6).

Sensitivity analysis.
A sensitivity analysis was conducted to test whether the meta-analysis results were robust. This inquiry into the sensitivity of the meta-analysis of IFN-γ levels in severe and uncomplicated malaria demonstrated that patients with severe malaria had higher mean IFN-γ levels than those with uncomplicated malaria when the leave-one-out method was applied (p < 0.05, Fig. 7), indicating that the results of the metaanalysis were robust. The sensitivity analysis between studies that reported mean/standard deviation and median/range and studies that reported mean without standard deviation was performed. Results showed that patients with severe malaria had higher mean IFN-γ levels than those with uncomplicated malaria (p < 0.001, MD: 20.12 pg/mL, 95% CI 9.56-30.69 pg/mL, I 2 : 98.39%, 12 studies/13 study sites, 598 severe cases/1020 uncomplicated cases, Supplementary Fig. S3). The sensitivity analysis between studies that reported mean/ standard deviation and those with median/range of IFN-γ levels in patients with severe and uncomplicated malaria was conducted using the subgroup analysis. Results indicated that patients with severe malaria had   Supplementary Fig. S4).
Publication bias. The publication bias of the effect estimates among the included studies was assessed by visualisation of funnel plot symmetry and Egger's test for small-study effect. In the meta-analysis between IFN-γ levels between patients with severe and those with uncomplicated malaria, asymmetry of the funnel plot was suspected (Fig. 8), and Egger's test exhibited a small-study effect (p < 0.001), showing that publication bias was discovered. The trim-and-fill method was applied to adjust the pooled effect estimate, and the rresults showed that the pooled MD of IFN-γ levels between patients with severe and uncomplicated malaria after adjusting for publication bias was 13.634 pg/mL (95% CI 6.979-20.29 pg/mL).

Discussion
The main feature of this study was the comparison of IFN-γ levels between patients with severe and those with uncomplicated malaria. The results of the qualitative synthesis demonstrated that most studies that investigated IFN-γ levels in both patients with severe and uncomplicated malaria revealed no statistically significant IFN-γ levels between the two clinical outcomes. Meanwhile, few studies have reported that IFN-γ levels were significantly higher in patients with severe malaria than those in patients with uncomplicated malaria. For quantitative synthesis by meta-analysis, the higher IFN-γ levels were observed in patients with severe malaria than those in   www.nature.com/scientificreports/ patients with uncomplicated malaria. Although a high degree of heterogeneity of the outcome existed, the metaanalysis results implied that IFN-γ levels were associated with malaria severity, i.e. increased levels positively correlated with increased severity. IFN-γ levels related to malaria severity were observed by in vitro studies [52][53][54] . Nevertheless, a study on children with malaria argued that reduced IFN-γ levels were associated with malaria severity 55 . A discrepancy in IFN-γ levels and malaria severity between studies might be because of differences in the different participants enrolled in each as suggested previously 20 . Various severe complications among patients, such as anaemia, parasitemia levels or cerebral malaria, might cause the differences in the MD in IFN-γ levels between severe and uncomplicated malaria. A previous study indicated that IFN-γ production was associated with the reduced prevalence of anaemia caused by P. falciparum; hence, IFN-γ was suggested to be an immunity-based protection against severe malarial anaemia 17 . A previous study found that IFN-γ levels were negatively associated with parasitemia, suggesting that this cytokine has antiparasitic effects 46 . Increased IFN-γ levels were also related to high malarial parasitemia 52 . This association may be due to malarial parasitemia caused by the production of IFN-γ by immune cells 56 . In comparison to severe malarial anaemia, patients with cerebral malaria showed higher levels of proinflammatory/Th1 cytokines 19 , indicating that the production of pro-inflammatory cytokines in severe malaria is poorly regulated 21 .
Although the meta-analysis results exhibited higher mean IFN-γ levels in severe malaria than in uncomplicated malaria, the degree of heterogeneity among studies included in the meta-analysis was extremely high. The meta-regression and subgroup analyses of study design, continents, age groups, and techniques for IFN-γ measurement proposed that these parameters were sources of heterogeneity in the outcome. Considering study design as a source of heterogeneity, prospective observational and case-control studies showed higher mean IFN-γ levels were found in patients with severe malaria compared to those with uncomplicated malaria. Meanwhile, the prospective cohort studies indicated no difference. These results might be because only two prospective cohort studies were included in the subgroup analysis, which might bias the results of the subgroup analysis. Considering the continent as another source of heterogeneity, studies conducted in Asia showed higher mean IFN-γ   www.nature.com/scientificreports/ levels in patients with severe malaria than those with uncomplicated malaria. However, there was no difference in mean IFN-γ levels among studies performed in Africa, indicating that the different populations investigated may have had various immune responses to malaria. In Africa, where falciparum malaria is endemic, the populations are more exposed to infections; hence, they may have acquired immune responses against malaria infections or severity 57-60 . Therefore, it is possible that both patients with severe and uncomplicated malaria showed comparable cytokine responses, which causes non-statistical significance between groups in this meta-analysis. In Asia, where falciparum malaria is less endemic, most populations are less exposed to infections; hence, they are non-immune or semi-immune responses to malaria infections 61 . Therefore, patients with severe malaria in Asian countries might develop a stronger immune response to the infections than those with uncomplicated malaria. For example, Singotamu et al. 24 in India indicated that P. vivax infections demonstrated very high mean IFN-γ levels (619.7 pg/mL) in patients with severe malaria compared with uncomplicated malaria (177.3 pg/ mL). Additional background histories or co-occurrence with pathogens other than malarial ones in various areas might manifest different immune responses in patients with malaria. The histories of the individuals studied may add a layer of complexity to the findings on the immune response to malaria 6 . Considering age groups as another source of heterogeneity in the outcome, no differences in mean IFN-γ levels in children with severe malaria and those with uncomplicated malaria, but adults with severe malaria Heterogeneity: τ 2 = 0.00, I 2 = .%, H 2 = .
Test of θi = θj: Q(14) = 1433.00, p = 0.00 Test of group differences: Qb (4)   www.nature.com/scientificreports/ showed higher mean IFN-γ levels than those with uncomplicated malaria. This result might be explained by the fact that studies enrolling children with severe malaria were conducted in Africa, where malaria is endemic. Meanwhile, studies enrolling adults with severe malaria were conducted in Asia, where malaria is less endemic. Across sub-Saharan Africa, where the disease is hyper-endemic, most people are almost continuously infected with P. falciparum. Most infected adults rarely experience severe disease because of the acquired immunity against the infection. In areas where malaria is less endemic, such as Asia, a higher risk of severe disease is frequently observed among adults than children as adults develop a stronger immune response to the infection, but infants and children occasionally do not. This reason explained the possible cause of high cytokine response, including IFN-γ response to the infection. Considering the techniques for measuring IFN-γ as another source of heterogeneity of the outcome, studies using ELISA exhibited higher mean IFN-γ levels in patients with severe malaria than in those with uncomplicated malaria. However, studies using bead-based assays indicated no differences in mean IFN-γ levels between patients with severe and those with uncomplicated malaria. Multiplex bead-based assays provide the means to simultaneously measure multiple proteins in a single reaction compared to ELISA, which measures a single protein in a cone reaction 62 . A study comparing the overall performance of the two methods for cytokine profiles demonstrated that the ELISA and bead-based assays yielded similar results 63 . Notably, ELISA was more sensitive in the low concentration range of the standard curve, whereas bead-based assays could detect higher protein concentrations 63 . Another study that measured IFN-γ levels using both techniques indicated comparable detection of plasma IFN-γ, IL-4, IL-10, and TNF-α, but the ELISA missed a cytokine such as IL-5 64 . Therefore, no study yet warrants the difference in performance between the two techniques for detecting IFN-γ in the blood. However, the results of the subgroup meta-analysis might guide further studies to examine the difference in the real performance of these techniques.
IFN-γ contributes to the activation and differentiation of B lymphocytes, T lymphocytes and macrophages [65][66][67] . The IFN-gamma receptor (R) locations were in lymphoid organs such as the B-cell areas of lymph nodes, spleen, tonsils, and in epithelial tissues of the intestinal system, lung, and endometrial mucosa cells 68 . Studies in mice models indicated that treatment of mice infected with blood-stage P. berghei by anti-IFN-γ antibody failed to control the infection parasites 69,70 . Additionally, delayed parasite elimination was found among IFN-γ-deficient   [71][72][73] . During Plasmodium infection, γδ T-cells that express CD40 ligand produce IFN-γ in response to infection by enhancement of dendritic cell activation to remove malaria parasites 74 . In the pathogenesis of severe malaria, many studies have indicated that IFN-γ is vital for developing severe malaria, particularly cerebral malaria, by affecting endothelial integrity 70,72,75,76 . During cerebral malaria, IFN-producing CD8 + T-cells are recruited to the brain and cause cerebral pathology by destroying the blood-brain barrier in perforin-and granzyme-dependent manner 77,78 . IFN-γ production is modulated by several cytokines such as IL-12 and IL-18 or broadly reactive antigen receptors 79 .    www.nature.com/scientificreports/ In the study by Wroczyńska et al. 25 , increased IFN-γ accompanied by increased IL-18 levels were observed in patients with severe malaria, indicating that excessive production of both cytokines is associated with severe malaria infections. IL-18-dependent IFN-γ overproduction was reported to relate to decreased IL-12 levels 25 . Therefore, these data suggested that severe malaria is associated with increased IFN-γ and decreased IL-12 levels, indicating the occurrence of immunoregulation in resolving malaria infection 25 . Furthermore, reduced IL-12 levels were associated with suppression of Th1 cytokine activation by NK cells or CD8 cells 56 . The previous studies also showed that IFN-γ was associated with IL-10 and IL-6, indicating a balance between these cytokines 23,80 . IFN-γ and IL-10 were markedly increased in patients with severe malaria 48,81 . This study had some limitations. First, the degree of heterogeneity was extreme in the meta-analysis. Although meta-regression and subgroup analyses were conducted to identify the source(s) of heterogeneity, the heterogeneity remained in the subgroup analysis, showing that other factors confound the association between IFN-γ levels and malaria severity. Second, publication bias among the studies included in the meta-analysis was noted. Therefore, the pooled effect estimate (MD of IFN-γ levels) after applying the trim-and-fill method should be considered.

Conclusion
In conclusion, patients with severe malaria present higher IFN-γ levels than those with uncomplicated malaria, although the heterogeneity of the outcomes is yet to be elucidated. To confirm whether alteration in IFN-γ levels of patients with malaria may indicate disease severity and/or poor prognosis, further studies are warranted.

Data availability
All data generated or analyzed during this study are included in this published article and its supplementary information files.