The Role of Eosinophils, Eosinophil-Related Cytokines and AI in Predicting Immunotherapy Efficacy in NSCLC

Biomolecules 2025 AI 6 Explanations View Original
Original Paper (PDF)

Unable to display PDF. Download it here or view on PMC.

Plain-English Explanations
Page 1
Eosinophils as Emerging Biomarkers of Immunotherapy Response in Lung Cancer

The biomarker gap in immunotherapy Immunotherapy with immune checkpoint inhibitors (ICIs) has transformed NSCLC treatment. But current biomarkers - PD-L1 expression, microsatellite instability (MSI), and tumor mutational burden (TMB) - are imperfect predictors. Many patients with high PD-L1 still do not respond, and some with low PD-L1 do. New biomarkers are urgently needed.

Eosinophils as candidates Eosinophils are white blood cells primarily known for their role in allergic diseases and parasitic infections. Emerging evidence suggests they play important roles in the tumor microenvironment and that their levels in blood (absolute eosinophil count, AEC) and tumor tissue (tumor-associated tissue eosinophilia, TATE) correlate with immunotherapy outcomes in NSCLC.

Key cytokines: IL-31 and IL-33 Two cytokines - IL-33 (an alarmin that activates eosinophils and other immune cells) and IL-31 (a T-cell-derived cytokine that regulates inflammation) - emerge as important modulators of eosinophil activity in the tumor microenvironment. Their roles in NSCLC immunotherapy response are reviewed here.

The AI angle The review argues that artificial intelligence - particularly machine learning applied to large multi-omic and clinical datasets - is needed to disentangle the complex, context-dependent roles of eosinophils in NSCLC and to develop validated eosinophil-based biomarker panels for clinical use.

TL;DR: This review summarizes evidence that eosinophil counts (AEC) and tumor tissue eosinophilia (TATE) correlate with immunotherapy outcomes in NSCLC, explores the roles of IL-31 and IL-33 in eosinophil activation, and proposes AI-driven analysis to translate these findings into clinical biomarkers.
Pages 2-3
How Eosinophils Develop and What They Do in Normal and Cancer Biology

Eosinophil differentiation Eosinophils develop from multipotent bone marrow progenitors through a series of differentiation steps driven by transcription factors (GATA-1, GATA-2, c/EBP-alpha) and cytokines (IL-3, GM-CSF, IL-5). IL-5 is particularly important and relatively specific to the eosinophil lineage; its overexpression in mouse models causes eosinophilia.

Eosinophil granule proteins Mature eosinophils contain granules packed with cytotoxic proteins: major basic protein (MBP), eosinophil cationic protein (ECP), eosinophil peroxidase (EPO), and eosinophil-derived neurotoxin (EDN). These proteins are released upon activation and can directly damage cells through oxidative mechanisms - the basis of eosinophil anticancer activity.

Dual role in tumors Eosinophils display paradoxical roles in cancer. Anti-tumor activities include promoting CD8+ T-cell infiltration, activating NK cells, inducing M1 macrophage polarization, releasing interferon-gamma, and directly killing cancer cells through degranulation. Pro-tumor activities include stimulating angiogenesis, releasing metalloproteinases (MMP-9), and promoting M2 macrophage polarization through TNF-alpha.

Context determines the outcome Whether eosinophils promote or suppress tumor growth depends on the cancer type, local cytokine environment, and the activation state of eosinophils. In melanoma they are predominantly anti-tumor; in cervical carcinoma and Hodgkin lymphoma they have been associated with worse prognosis. In NSCLC, the picture is emerging as predominantly favorable.

TL;DR: Eosinophils are pleiotropic immune cells that can kill cancer cells directly through cytotoxic granule proteins and indirectly by recruiting CD8+ T cells and NK cells, but also have pro-tumor effects in some contexts through angiogenesis promotion and M2 macrophage polarization.
Pages 4-5
Two Cytokines that Regulate Eosinophils and the Tumor Immune Environment

IL-33: an alarmin with dual roles IL-33 is an alarmin cytokine released by epithelial and tumor cells upon cellular stress or damage. It binds ST2 receptor and activates Th2 cells, mast cells, ILC2s, and eosinophils - promoting IL-4, IL-5, and IL-13 release. IL-33 also directly activates CD8+ T cells and NK cells for anti-tumor killing, and promotes the formation of immunogenic dendritic cells (cDC1s) that stimulate anti-cancer T-cell responses.

IL-33's paradoxical cancer effects In IL-33-knockout mice, tumor growth is faster, suggesting baseline IL-33 suppresses tumor development. But overexpression of IL-33 also improves immune responses to tumors. The dual role depends on which cell populations are activated: activating eosinophils and CD8+ T cells is anti-tumor, while polarizing macrophages to M2 and expanding regulatory T cells is pro-tumor.

IL-31: a T-cell cytokine with emerging roles IL-31 is produced by CD4+ T cells and activates the JAK/STAT, PI3K/AKT, and MAPK pathways, regulating inflammation, cell proliferation, and tissue remodeling. In breast cancer models, IL-31 injection increased CD8+ T-cell infiltration and improved survival. In gastric cancer, IL-31 levels are reduced. In endometrial cancer, higher IL-31 levels correlate with immune cell infiltration and better outcomes.

Combined biomarker potential Levels of IL-31 and IL-33 correlate with staging, invasion, and metastasis in several cancer types, and their sensitivity and specificity as tumor markers surpass traditional markers (CEA, CA-125, CA19-9) when combined with clinical parameters. Their roles in NSCLC specifically under ICI treatment are insufficiently studied and represent an important research priority.

TL;DR: IL-33 and IL-31 regulate eosinophil activity and broader immune cell function in the tumor microenvironment, with IL-33 showing paradoxical pro- and anti-tumor effects depending on context, and both cytokines demonstrating early potential as cancer biomarkers superior to traditional markers.
Pages 8-10
Clinical Evidence Linking Eosinophil Counts to Immunotherapy Outcomes in Lung Cancer

AEC and survival on ICIs Multiple retrospective studies in NSCLC patients treated with anti-PD-1/PD-L1 agents have reported that higher absolute eosinophil count (AEC) correlates with better outcomes. One study of 121 patients found that AEC above 500 cells/uL correlated with significantly better OS (26.6 vs. 9.5 months) and PFS (13.8 vs. 4.6 months). Another study showed median OS of 667 days in the 100-500 cells/uL group vs. 339 days in the under 100 cells/uL group.

AEC dynamics and disease control A study of 151 NSCLC patients found that patients with increased eosinophil counts 4 weeks after ICI initiation had higher disease control rates and median OS of 674 vs. 234 days compared to those without eosinophil increases. This suggests that early eosinophil rise during treatment may be a marker of favorable immune activation in response to ICI.

TATE as a tissue biomarker Tumor-associated tissue eosinophilia (TATE) - the presence of eosinophils within tumor tissue - has been linked to improved overall survival in meta-analyses of solid tumors, inversely correlated with lymph node metastasis, and is a prognostic indicator across multiple cancer types. Its specific prognostic value in NSCLC remains under investigation.

Eosinophilia and adverse events The relationship between eosinophils and immunotherapy toxicity is bidirectional: eosinophilia correlates with favorable responses but very high AEC (greater than 500 cells/uL) is associated with a higher frequency of immune-related adverse events (irAEs). This suggests eosinophil levels reflect systemic immune activation that benefits anti-tumor response but can also cause autoimmune complications.

TL;DR: Clinical evidence from multiple studies shows that AEC above 150-500 cells/uL correlates with improved OS and PFS in NSCLC patients receiving ICIs, while eosinophil increases during treatment may serve as an early marker of immunotherapy benefit - though very high levels also predict toxicity.
Pages 11-12
How Artificial Intelligence Can Unlock the Biomarker Potential of Eosinophils

The complexity problem Eosinophil biology in cancer is highly context-dependent, with roles that vary by cancer type, treatment regimen, patient demographics, comorbidities, and the broader immune context. The complex interactions between eosinophils, IL-33, IL-31, PD-L1, CD8+ T cells, and other biomarkers cannot be adequately modeled with traditional univariate or multivariate statistical approaches.

AI for big data integration Machine learning and deep learning models can integrate large datasets combining eosinophil counts, cytokine levels, tumor genomics, imaging features, and clinical outcomes to discover biomarker patterns that no single variable reveals. AI can handle high-dimensional data, capture nonlinear interactions, and identify patient subgroups with distinct eosinophil-mediated response patterns.

Predicting both response and toxicity AI models that simultaneously predict ICI efficacy and toxicity risk based on eosinophil levels and other biomarkers could enable truly personalized treatment planning - identifying patients likely to benefit from ICIs while flagging those at high risk for eosinophilia-associated adverse events. This dual optimization is beyond the capability of standard biomarker thresholds.

Histopathology and digital pathology AI-powered digital pathology can quantify TATE from routine tumor biopsy slides automatically and consistently, overcoming the limitation that eosinophils are sparse in tissue and difficult to count manually. Automated eosinophil density mapping could make TATE a scalable, reproducible biomarker for clinical use.

TL;DR: AI is needed to model the complex, context-dependent relationship between eosinophil levels, cytokines, and ICI outcomes in NSCLC; machine learning offers the ability to integrate multidimensional data and develop validated biomarker panels for clinical use.
Pages 12-13
Knowledge Gaps and the Research Needed to Translate Eosinophil Biology to Clinical Practice

Predominantly retrospective evidence Most clinical data on eosinophils and ICI outcomes in NSCLC comes from retrospective analyses with small cohorts, variable eosinophil cutoffs, and heterogeneous patient populations. Prospective studies with pre-specified eosinophil biomarker endpoints are needed to establish consistent thresholds and validate prognostic utility.

Mechanistic gaps in NSCLC While eosinophil biology in other tumors (melanoma, colon cancer) is relatively well described, the mechanisms by which eosinophils interact with NSCLC cells and the ICI-activated immune environment are incompletely understood. Mouse models specific to lung cancer are needed to establish causal relationships.

Standardization challenges AEC measurement varies with timing (pre-treatment vs. during treatment), analyzer platforms, and laboratory reference ranges across institutions. TATE quantification lacks standardized scoring criteria. Before eosinophil biomarkers can enter clinical practice, standardized measurement protocols must be established and validated.

Future directions Prospective multicenter NSCLC cohort studies with serial AEC measurements, paired TATE quantification, and systematic cytokine profiling (IL-33, IL-31) are the primary research need. AI-powered analysis of these datasets could identify composite biomarker signatures with superior predictive accuracy. Development of standardized clinical assays and regulatory approval pathways will then be needed for clinical translation.

TL;DR: The field needs prospective studies with standardized eosinophil measurement protocols to establish validated thresholds; AI-powered integration of eosinophil counts, cytokine levels, and tumor features across large NSCLC cohorts represents the pathway to clinical translation.
Citation: Open Access, 2025. Available at: PMC12024677.