Low eosinophils and their dynamic as a predictor of death in patients with infections: a systematic review and meta-analysis of cohort studies

1. Introduction

Eosinophils represent 0–7% of the peripheral blood cells, being active components of both innate and adaptive immune response that produce and store cationic proteins, interleukins and other components, being involved in protecting the host against several types of infections caused by parasites, viruses, fungi and bacteria [1]. In addition, eosinophils could potentially take on novel roles, such as neoplasm surveillance, contributing to tissue repair and regeneration [2].

Until recently, their clinical significance was primarily related to increase in their absolute value while their decrease generally went unnoticed. A low value of absolute eosinophil count (AEC) may fall within the laboratory normal range, as there is still a lack of consensus regarding the optimal cut-off for defining eosinopenia. Potential causes for eosinophil number decrease had been described secondary to medications, associated with infections, stress response, increased cortisol levels, or critical illness [3].

A low eosinophil count (<50 cells/mm³) effectively discriminates between patients with and without bacterial infections [4], presenting 80% sensitivity and 91% specificity [5]. Additionally, eosinophil dynamics might offer a superior indicator for early antibiotic response compared to C reactive protein [6].

While several studies have examined the prognostic value of eosinophils, it wasn’t until the COVID-19 pandemic that a significant interest emerged in identifying precise prediction factors for patient outcomes [7]. This focus particularly extended to peripheral eosinophils, highlighting their potential for outcome assessment in patients with viral infections [8]. However, their exact value as mortality markers remains unclear due to conflicting literature. Previous attempts to explored the association between eosinopenia and mortality in COVID-19 [9,10] and septic patients [11,12] focused solely on admission eosinophil count. We aim to explore this topic from a novel and more comprehensive perspective, emphasizing for the first time eosinophil dynamics as a prognostic factor.

We conducted a systematic review (SR) of cohort studies and carried out a meta-analysis (MA) to investigate the potential impact of low eosinophil counts (eosinopenia) on the survival rates of patients with viral, bacterial, and fungal infections. Secondary objectives included assessing the influence of sustained low eosinophil levels and eosinophil recovery during hospitalization on patient outcomes.

2. Methods

The systematic review adheres to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines and to the Checklist for critical Appraisal and data extraction for systematic Reviews of prediction Modelling Studies (CHARMS checklist). The protocol was registered on PROSPERO (ID: CRD42023388226).

3. Results

From 15,066 papers screened, we included 151 papers (137 original articles, 11 conference papers, and 3 article abstracts). The PRISMA flowchart of the study is presented in Figure 1, while the reference list of the included studies and the list of excluded papers at full text screening are provided in Supplemental Section 17 and 18.

We identified 107 papers related to COVID-19 patients, 14 papers on sepsis, nine on C. difficile, and six on AECOPD. In addition, isolated papers assessed eosinophils in various infections: pneumonia (seven), bloodstream infections (two), influenza (two), severe fever with thrombocytopenia syndrome virus (SFTS) (two), mixed types of infections (three), and Talaromyces marneffei (one).

4. Risk of bias and publication bias

In terms of methodological quality, regardless of the infectious pathology, more than 50% of the overall risk of bias for each infection group was categorized as either high or moderate (Supplemental Section 10; Figure 10.1–10.11). This assessment was primarily influenced by concerns related to confounding, participation, statistical analysis, and reporting domains. Due to the short duration of the studies, there was a low risk of bias related to attrition.

Visual interpretation of the funnel plots suggested low risk of publication bias among the analyzed papers (Supplemental Section 14, Figure 14.1–14.3).

5. Discussion

Although animal studies showed that eosinophils are not vital to our survival [32], in patients with infections, we found evidence that low peripheral eosinophils might be associated with in-hospital mortality.

Typically, infection-induced eosinopenia can be attributed to several mechanisms. Firstly, heightened sympatho-adrenal activity results in elevated stress-induced cortisol levels, which can suppress bone marrow function [33]. Furthermore, the baseline inflammatory state and cytokine imbalance facilitates the migration of eosinophils to the inflamed tissues [34], redirecting them from circulation. In SARS–CoV-2 infection, the increased viral load depletes CD8 T cells, reducing interleukin-5 which is critical for eosinophil functions [35], while also triggering eosinophil granule protein consumption [36]. Eosinophil levels mirror the severity of immune dysregulation in COVID-19, correlating with various inflammatory markers (IL2R, IL6, IL8, TNF-α, CRP, and procalcitonin) [37]. In C. dfficile infection, the mechanism connecting peripheral eosinopenia with mortality remains unclear, but mouse models indicate that C. difficile binary toxin can induce circulating eosinophils apoptosis by binding to surface Toll-like receptor 2 [38].

Overall, there was a trend towards lower admission eosinophil values in the deceased. Discrepancies in the statistical significance, in patients with COVID-19 and other bacterial infections, might be explained by variations in races, ethnicity, treatment practices, vitamin D deficiency [39], time from symptom onset to eosinophil measure, viral genomic differences, and vaccination status [40], as well as COVID-19 infection severity [24]. In addition, males tend to have higher eosinophil levels than females [41]. With advancing age, immunosenescence contributes to altered eosinophil homeostasis and impaired function, which may compromise the immune responses, but it is not associated with a reduced eosinophil count [41,42].

In general, mortality was higher among patients with lower eosinophil values, although the certainty of this evidence was low for C. difficile infection and very low for COVID-19.

In vivo the half-life of blood eosinophil’s ranges between 8 and 18 h [43], while their tissue lifespan increases to two to 5 days in physiological conditions and reaches approximately 2 weeks in inflammatory states [1,44], changes that might explain eosinophil recovery during the course of the disease.

In addition, the gradual increase (recovery) of eosinophils was linked to survival, whereas persistent eosinopenia increased the mortality risk. Although the mechanisms linking eosinophil recovery to improved clinical outcomes remain unclear, this trend may be explained by the gradual resolution of factors previously implicated in eosinopenia, such as acute stress responses (like elevated corticosteroids), or systemic inflammation. In studies assessing eosinophil dynamics included in our SR, a similar pattern of immune cell recovery was observed (Supplemental Section 19; Table 1). Lymphocyte counts remained low in deceased patients but recovered or stabilized in survivors. Neutrophils were persistently elevated in non-survivors, reflecting ongoing inflammation, while survivors showed stable or mildly increased levels. Monocyte counts were generally lower in deceased patients early on, with some later increases, whereas survivors maintained stable or moderately elevated levels. Limited data indicate basophils were suppressed in deceased patients but more stable or slightly higher in survivors, suggesting better immune regulation. The stepwise increase in eosinophil counts observed in survivors may reflect the progressive restoration of immune homeostasis and resolution of immune dysregulation. The recovery of eosinopenia, alongside similar upward trends in other white blood cell lines among survivors, suggests that eosinophil restoration may reflect not an isolated phenomenon but a component of broader immune system reconstitution following earlier immune dysregulation.

Based on our analysis, it is expected that eosinophil counts will increase within the initial 2–5 days, although its important to note that delays in their recovery might arise depending on the clinical course of the infection and the different treatments received, including corticotherapy. According to Merino et al. [45], a count exceeding 210 cells/mm³ might be among the numerous criteria considered when deciding ICU discharge for patients with severe sepsis and septic shock. Additionally, notable rises in eosinophil count during hospitalization may stem from antibiotic-induced allergy or the unmasking of preexisting eosinophilia as sepsis resolves [46].

Given the increasing clinical focus on personalized medicine, monitoring eosinophil dynamics could serve as a practical, cost-effective tool in guiding immune-targeted therapies especially valuable in low-resource settings where access to more advanced tools may be limited. By tracking eosinophil levels, clinicians may be able to tailor treatment approaches, particularly in infections where immune modulation is key, making it highly relevant for both immunologists and clinicians seeking to optimize patient care in real-time.

This was the first SR and MA to evaluate admission eosinophil counts and lower different eosinophil cut-offs in relation with mortality focusing on eosinophil dynamic changes during admission in various infectious diseases. Our analysis uniquely captures how eosinophil recovery trajectories during hospitalization correlate with patient outcomes, suggesting new insights into their role in immune response modulation and infection prognosis.

A key strength of our systematic review lies in its comprehensive approach to literature synthesis, encompassing a thorough search across published original articles and conference papers, allowing a nuanced evaluation of eosinophil’s as a prognostic marker for mortality across a broad spectrum of infectious diseases.

However, our analysis encountered significant limitations, such as diverse outcome measurement methods (ORs, HRs, and AUROCs often lacking patient outcome numbers), increased study heterogeneity, suboptimal methodological quality, and inconsistent reporting of eosinophil counts – most commonly presented as medians with interquartile ranges, which limited data synthesis. Most studies exhibited a high risk of bias, primarily due to misalignment with our systematic review’s objectives. However, confounding was less impactful, as eosinopenia was assessed purely as a prognostic, not causal, factor [47].

Our meta-analyses showed consistently high heterogeneity (I² > 90%), which undermines the validity of the pooled effect estimates. This variability is attributable to differences in infection types (viral vs. bacterial), population characteristics (such as age, sex, and comorbidities), clinical settings (ward vs. ICU), and study designs. Other contributing factors include varying thresholds for eosinopenia and differences in statistical reporting formats (mean vs. median requiring data conversion). While subgroup and sensitivity analyses were performed to explore these sources, heterogeneity remained high. Meta-regression was considered; however, it was not feasible due to insufficient reporting of key covariates and widespread heterogeneity in data presentation. This limitation should be considered when interpreting the pooled results. Although the use of Google Translate might represent a potential limitation, by introducing inaccuracies or misinterpret nuances, only 3% of the articles required translation.

Despite the considerable heterogeneity across studies evaluating static eosinophil measures, this systematic review provides a comprehensive synthesis of eosinophil dynamics and their prognostic significance across a broad spectrum of infectious diseases. While variability remains high, the consistent trends identified emphasize the need for further research to establish standardized eosinophil cut-offs, assess daily changes in cell dynamics, and define recovery thresholds across diverse hospital settings. These efforts are crucial for assessing the clinical benefits of this widely available marker.

6. Conclusions

Our analysis indicates a low certainty of evidence linking absolute eosinopenia (0 cells/mm³) with higher mortality in Clostridioides difficile infections. In COVID-19, the evidence remains uncertain, though a trend toward higher mortality with eosinopenia was observed. Deceased patients consistently showed lower eosinophil levels than survivors, suggesting that monitoring eosinophil recovery, rather than static levels, could serve as a cost-effective and reliable predictor of patient outcomes. Further research is essential to validate these findings.

Supplementary Material

Statement

During the preparation of this work the authors used ChatGPT 4.0 in order to improve readability. After using this tool/service, the authors reviewed and edited the content as needed and take full responsibility for the content of the publication.

Disclosure statement

The authors have no conflicts of interest to declare.

Primary Funding Source

None. PROSPERO (CRD42023388226)

Data availability statement

The authors confirm that the data supporting the findings of this study are available within the article and its supplementary materials. Derived data supporting the findings of this study are available from the corresponding author LP on request.