Overall, 417 participants with AFI presenting at the outpatient clinic were screened for eligibility and 415 children (227 female, 54.7%; 188 male, 45.3%) were included in this analysis (Fig. 1). Median (IQR) age was 6.0 (4.0–10.0) years and most children (n = 202, 48.7%) were in the age group between 5–11 years. 86 of 415 participants (20.7%) presented with fever, and the median (IQR) fever duration was 3.0 (2.0–4.0) days. In total, 174 of 415 (41.9%) participants tested positive for malaria by microscopy, of whom 9 (5.2%) had concomitant bacterial and 39 (22.4%) non-bacterial infections. Infection with Plasmodium falciparum (158, 90.8%) was most prevalent. Among 241 (58.1%) M[-] patients, 23 (9.5%) had bacterial and 54 (22.4%) non-bacterial infections. Overall, no etiologic cause of AFI was identified in 164/415 (39.5%) participants. Rapid diagnostic tests for Human Immunodeficiency Virus (HIV) yielded positive results in 6 of 416 (1.4%) children, equally distributed between both M[+] and M[−] children. Children diagnosed with malaria were significantly older (7.0 [4.0–11.0] years vs. 6.0 [4.0–10.0] years, p = 0.037) than M[−] peers (Table 1).
Demographic and clinical information on specific analysis groups shown in Fig. 1 can be found in the Supplementary Material (Supplementary Table 1).
Malaria is associated with increased odds for anemia
Among 297 of 415 participants (71.6%), in whom full blood count was available, anemia was highly prevalent. Of those, 222 were anemic, resulting in an estimated point prevalence of 74.7% (95% CI 69.5 –79.4) overall. Most anemic children presented with moderate (n = 146, 49.2%) followed by mild (n = 54, 18.2%) and severe (n = 15, 5.0%) anemia. The estimated point prevalence of anemia according to malaria status was 82.9% (92/111, 95% CI 74.6–89.4) in M[+] and 69.9% (130/186, 95% CI 62.8–76.4) in M[−] children.
M[+] children had a significantly higher odds of anemia compared to M[-] children (ORunadjusted 2.09, 95% CI 1.16–3.74). There were neither sex nor age specific differences regarding the odds of anemia with respect to malaria status. (Supplementary Table 2) Moreover, there was a moderate association between anemia severity and malaria status (V = 0.32, X2(3, N = 297) = 29.98, p < 0.001). Post-hoc comparison of malaria case proportions showed significant differences across all levels of anemia severity, with malaria prevalence increasing as anemia severity worsened. This trend was further supported by decreasing odds of a negative TBS with advancing anemia severity (no anemia: OR = 2.09, 95% CI 1.16–3.88; mild anemia: OR = 3.12, 95% CI 1.46–7.29; moderate anemia: OR = 0.46, 95% CI 0.27–0.76; severe anemia: OR = 0.14, 95% CI 0.02–0.52; Fig. 2a). Moreover, M[+] children had a significantly higher median red blood cell distribution width (RDW) compared to M[−] and, although not statistically significant, tended to be more microcytic and hypochromic (Supplementary Table 3).
Frequency distributions of (a) severity of anemia, (b) type of iron deficiency, and (c) anemia classified according to TSAT and ferritin concentrations in malaria positive (tbs pos) and malaria negative (tbs neg) patients. Graph (d) shows a matrix layout for number of patients that group according to different iron marker cut-offs. Black dots and connecting lines indicate possible combinations of biomarker cut-offs meeting the definitions of AI, AI/IDA, IDA or not. Abbreviation: ID iron deficiency, IR iron replete, tbs thick blood smear, TSAT transferrin saturation, AI anemia of inflammation, AI/IDA anemia of inflammation with possible iron-deficiency anemia, IDA iron-deficiency anemia.
Iron deficiency and fever
Overall, among 310 of 415 patients (74.7%) with available serum iron parameters, 72 (23.2%) had absolute ID (ferritin <100 µg/L), 107 (34.5%) functional ID (ferritin ≥ 100 µg/L), and 131 (42.3%) had normal iron status. When assessing the distribution of iron deficiency types according to malaria status, Chi-squared test results revealed a significant association (V = 0.15, X2(2, N = 310) = 7.27, p = 0.026). Post-hoc testing showed a significantly greater proportion of children having absolute ID in the M[−] compared to the M[+] group (28.5% vs. 16.5%, p = 0.046, Fig. 2b). In a multivariate logistic regression model for ID, defined as transferrin saturation (TSAT) < 20%, stratified for sex and adjusted for age, body temperature (OR 1.88, 95% CI 1.39–2.54) and a duration of fever ≥ 3 days (OR 0.57, 95% CI 0.35–0.93) were significantly associated with ID (Supplementary Table 4).
Anemia of inflammation
In 197 of 415 patients (47.5%), hemoglobin concentration and parameters of iron homeostasis were available, and among them, 145 (73.6%) were anemic. Among anemic patients 53 (36.6%) had AI, 29 (20.0%) AI/IDA, and 11 (7.6%) IDA irrespective of cause of AFI. Another 52 cases (35.9%) were attributed to multifactorial genesis and showed overall the picture of microcytic normochromic anemia (median [IQR] mean corpuscular volume, MCV 76 [68–78] fl; mean corpuscular hemoglobin concentration, MCHC 316.5 [307.0–325.3] g/L) with median (IQR) ferritin levels of 244 (103–470) µg/L, transferrin levels of 2.39 (2.01–2.57) g/L, and TSAT of 27 (23–33) %. The frequency of AI was highest in the M[+] group followed by multifactorial anemia (Fig. 2c). In this multifactorial group, 17/52 (32.7%) cases exhibited transferrin concentrations below the lower reference level (set at 2.2 g/L) and ferritin ≥100 µg/L, while 12/52 (23.1%) children had normal transferrin levels and ferritin concentrations <100 µg/L (Fig. 2d).
Median red blood cell (RBC) counts, MCV, mean corpuscular hemoglobin (MCH), MCHC, and RDW differed significantly between patients classified with AI, AI/IDA, and IDA, yet hemoglobin concentrations and hematocrit did not vary significantly. AI patients had significantly higher median c-reactive protein (CRP) levels compared to other groups except for multifactorial anemia. When comparing hepcidin and soluble transferrin receptor (sTfR) levels among classification of anemia groups, no statistically significant differences were observed. Moreover, no specific cytokine profiles were found to be associated with a specific type of anemia (Supplementary Table 5).
Iron metabolism and immune activation during malaria infection
Among children in the iron deficiency analysis group, significantly higher median plasma ferritin (244 [120–379] µg/L vs. 114 [60–230] µg/L, p < 0.001) and lower median plasma transferrin concentrations (2.45 [2.15–2.80] g/L vs. 2.65 [2.38–3.01] g/L, p < 0.001) were observed in M[+] compared to M[−] patients. A significant negative correlation between these iron parameters was observed within both groups (ρ(M[+]) = −0.55, 95% CI −0.67 to −0.39; ρ(M[−]) = −0.53, 95% CI −0.65 to −0.38).
When considering all participants with available hepcidin and sTfR levels, hepcidin concentrations were slightly higher in the M[+] group (36 ng/mL, IQR 29–79 ng/mL) compared to the M[−] group (33 ng/mL, IQR 27–45 ng/mL, p = 0.038). However, no statistically significant differences were observed in analyses limited to complete cases of iron markers or in subgroup analyses.
Malaria is not only associated with deranged iron metabolism but also alterations in circulating cytokine levels. Markers of immune activation were assessed in a subset of 54 children (iron status and cytokines analysis subgroup: n(M[ + ]) = 35, n(M[−] = 19) with either malaria infection or undetermined cause of AFI (Fig. 1), where both iron and cytokine levels were available. All evaluated parameters (interleukin (IL)−2, IL-6, IL-8, IL-10, tumor necrosis factor (TNF)-α, IFN-γ) were significantly higher in the M[+] group except for IL-4 (Supplementary Table 6). The cytokines exhibiting the most substantial differences between the M[+] and M[−] groups were IL-10 (r[U] = 0.71) and TNF-α (r[U] = 0.44), followed by IL-2 (r[U] = 0.34), IFN-γ (r[U] = 0.31), IL-8 (r[U] = 0.29) and IL-6 (r[U] = 0.28). The TNF-α: IL-10 ratio was significantly decreased in the M[+] compared to the M[−] group (r[U] = 0.56, p < 0.001).
Malaria parasite counts were significantly correlated with various cytokine concentrations, listed here by descending order in strength of association: IL-10 (ρ = 0.72, 95% CI 0.51–0.85), IL-6 (ρ = 0.63, 95% CI 0.37–0.79), IL-8 (ρ = 0.39, 95% CI 0.06–0.64), and TNF-α (ρ = 0.39, 95% CI 0.06–0.64) (Fig. 3a). Moreover, ferritin and CRP levels were also positively associated with malaria parasitemia (Fig. 3a).
Graph (a) shows results of M[+] children (n = 35), and graph (b) for M [−] children with unidentified cause of AFI (n = 19) group. Associations between variables were calculated using Spearman’s correlation coefficient and only statistically significant associations (punadjusted < 0.05) are shown. Colors range from red (negative association) to white (no association) to blue (positive association), as indicated in the scale bar. Abbreviations: Hct hematocrit, CRP c-reactive protein, IL interleukin, IFN-γ interferon-γ, TNF-α tumor necrosis factor α.
In the M[+] group (N = 35), both SI and TSAT showed a significant negative correlation with levels of IL-2 (SI: ρ = −0.52, 95% CI −0.73 to −0.23; TSAT: ρ = −0.42, 95% CI −0.66 to −0.1), IL-6 (SI: ρ = −0.46, 95% CI −0.69 to −0.14; TSAT: ρ = −0.34, 95% CI −0.61 to −0.01), and IL-10 (SI: ρ = −0.49, 95% CI −0.7 to −0.18; TSAT: ρ = −0.34, 95% CI −0.61 to −0.01). Yet in the M[−] group (N = 19), TSAT levels were positively correlated with levels of IL-4 (ρ = 0.57, 95% CI 0.15–0.81), IFN-ɣ (ρ = 0.71, 95% CI 0.37–0.88). Moreover, ferritin concentrations correlated positively with IFN-ɣ (ρ = 0.69, 95% CI 0.34–0.87), while transferrin exhibited an inverse relationship with IFN-ɣ (ρ = −0.46, 95% CI −0.76 to −0.01). (Fig. 3b) Overall, IL-4 levels correlated positively with ferritin (M[+]: ρ = 0.43, 95% CI 0.11–0.67; M[-]: ρ = 0.68, 95% 0.32–0.87) and negatively with transferrin (M[+]: ρ = −0.38, 95% CI −0.63 to −0.05; M[−]: ρ = −0.52, 95%CI −0.79 to −0.09) in both groups.
Iron status and immune activation in coinfections
In children of the anemia analysis group, median RBC count (4.34, [3.80–4.79] × 1012/L vs. 4.61, [4.27–5.02] × 1012/L, p = 0.013), median hemoglobin concentration (94 [86–108] g/L vs. 106.0 [100–113] g/L, p = 0.009), and median hematocrit (30.5 [27.5–34.1] vs. 33.4 [31.4–35.2], p = 0.005) were significantly lower, while median RDW (14.90, [13.9 –16.40] % vs. 14.00 [13.08–15.50]%, p = 0.005) was higher in children with malaria infection compared to those with B/NB infections. Interestingly, children with B/NB infections exhibited no significant variations in RDW irrespective the malaria status (Supplementary Table 7).
Among children in the iron status and coinfections analysis subgroup (N = 230/415, 55.4%, Fig. 1), children with malaria + B/NB showed significantly higher median [IQR] ferritin levels (286 [178–409] vs. 190 [64–310], p = 0.037, Fig. 4a) and an increased median [IQR] ferritin: transferrin ratio (97 [40–169] vs. 67 [25–118], p = 0.045, Fig. 4c) compared to those with only B/NB infection. Apart from ferritin, no statistically significant variations in single iron parameters were observed between groups with identified pathogens (Fig. 4b, d–f) (Supplementary Table 8).
Plasma concentrations of (a) ferritin, (b) transferrin, c ferritin: transferrin ratio, (d) TSAT, (e) hepcidin, and (f) sTfR) were determined in children with bacterial or non-bacterial infection (B/NB, red), Malaria without co-infections (Malaria, light blue), Malaria with bacterial or non-bacterial co-infections (Malaria + B/NB, green), and Undetermined cause of infection (Undetermined, dark blue). Black dots and lines indicate group median and group interquartile range, respectively. P values were determined using Kruskal–Wallis test followed by Mann–Whitney post hoc testing with Benjamini–Hochberg adjustment for multiple comparisons. Significance levels were defined as follows, ns: padj > 0.05; *: padj ≤ 0.05; **: padj ≤ 0.01; ***: padj ≤ 0.001; ****: padj ≤ 0.0001. Only statistically significant differences from the post-hoc analysis are shown. Scales in a–f are adjusted to accurately display datapoints up to the 95th percentile. Cohort-based reference ranges for ferritin (10.9–81.1 µg/L) and transferrin (2.23 to 3.56 g/L) were recently assessed in European children and were found to be sex and age-dependent.17 There are no established reference values currently available for the hepcidin and sTfR assays. Abbreviations: AFI acute febrile illness, B/NB bacterial/non-bacterial, TSAT transferrin saturation.
Cytokine levels were assessed in 78 of 415 children (18.8%; cytokine analysis group) with different causes of AFI. Both TNF-α and IL-10 levels were significantly higher in the Malaria + B/NB compared to the B/NB infection group (Fig. 5a, b). IL-10 levels significantly differed between groups with an identified pathogen (Fig. 5b). These differences in immune activation patterns among different infection groups were best represented by the TNF-α: IL-10 ratio (Fig. 5c). Other cytokines exhibited no notable trends between disease entities (Fig. 5d–f and Supplementary Table 9). TNF-α, IL-10, and IL-6 levels were widely distributed in the Malaria group.
Serum concentrations of a TNF-α, (b) IL-10, (c) TNF-α:IL-10 ratio, (d) IL-6, (e) IFN-γ, and f IL-4 were measured children with bacterial or non-bacterial infection (B/NB, red), malaria without co-infections (Malaria, light blue), malaria with bacterial or non-bacterial co-infections (Malaria + B/NB, green), and undetermined cause of infection (Undetermined, dark blue). Black dots and lines indicate group median and group interquartile range, respectively. P values were determined using Kruskal–Wallis test followed by Mann–Whitney post hoc testing with Benjamini–Hochberg adjustment for multiple comparisons. Significance levels were defined as follows, ns: padj > 0.05; *: padj ≤ 0.05; **: padj ≤ 0.01; ***: padj ≤ 0.001; ****: padj ≤ 0.0001. Only statistically significant differences from the post-hoc analysis are shown. Scales in a–f are adjusted to accurately display datapoints up to the 95th percentile. There are no established reference values currently available for these cytokine assays. Abbreviations: AFI acute febrile illness, B/NB bacterial/non-bacterial, IL interleukin, IFN-γ interferon-γ, TNF-α tumor necrosis factor α.
