Baseline IL-6, IL-8, IFN-ω, and perforin as prognostic biomarkers in immune checkpoint inhibitor-treated metastatic non-small cell lung cancer

Introduction

Background

Immune checkpoint inhibitors (ICIs) have transformed the treatment paradigm for advanced non-small cell lung cancer (NSCLC), yet durable benefit is achieved in only a subset of patients. Current predictive biomarkers, including programmed death ligand-1 (PD-L1) expression and tumor mutational burden (TMB), incompletely capture treatment response. PD-L1 expression alone is not a definitive predictor of immunotherapy benefit and remains spatially and temporally heterogeneous across the disease course (1). Similarly, while TMB has been investigated as a biomarker of ICI efficacy, the absence of standardized methodologies across platforms limits its clinical utility (2). Collectively, neither PD-L1 nor TMB adequately reflect the dynamic tumor microenvironment or intratumoral heterogeneity.

Cytokines and chemokines provide a more dynamic measure of systemic immune activation and inflammation and have shown promise as prognostic biomarkers in ICI-treated patients (3). Elevated interleukin-6 (IL-6) and interleukin-8 (IL-8) have been consistently associated with resistance to ICIs, whereas immune-stimulatory cytokines and cytotoxic effectors, including type I interferons and perforin correlate with improved antitumor responses (4-7). A recent meta-analysis demonstrated that elevated baseline IL-6 levels were associated with inferior progression-free survival (PFS) in NSCLC patients receiving programmed death 1 (PD-1) inhibitors (8). In a prospective study, Kang et al. (9) reported that patients with lower baseline IL-6 had significantly longer PFS (6.3 vs. 1.9 months) and improved overall survival (OS) (not reached vs. 7.4 months) compared with those with higher IL-6. Other studies have similarly shown that elevated IL-6 and IL-8 predict worse survival and reduced clinical benefit with ICIs (10,11). Mechanistically, IL-6 plays a context-dependent role, acting as a tumor suppressor in early carcinogenesis (12) but, in advanced disease, promoting proliferation, invasion, angiogenesis, and immune evasion through IL-6-STAT3 signaling.

Conversely, higher baseline perforin levels have been linked to improved PFS and OS in ICI-treated NSCLC (13). Increased baseline concentrations of type I interferons have also been associated with favorable outcomes in patients receiving chemoimmunotherapy, particularly in lung cancers. Among them, interferon-omega (IFN-ω), a member of the type I interferon family, has been shown to inhibit tumor growth by acting directly on tumor and immune cells (14). IFN-ω modulates the tumor microenvironment and enhances immune cell activation, underscoring its protective association with survival (14).

Rationale & knowledge gap

While individual cytokines have correlated with ICI outcomes, comprehensive analyses of broader cytokine signatures in prospective cohorts remain limited. Most prior studies have evaluated individual cytokines in isolation, offering limited knowledge around the role of integrated cytokine profiles in providing prognostic information beyond established tumor-based biomarkers. We hypothesized that baseline circulating cytokine profiles could provide robust prognostic information in NSCLC patients treated with ICIs.

Objective

To investigate whether baseline circulating cytokine profiles provide prognostic information in patients with advanced NSCLC receiving ICIs. Using a prospective single-institution observational cohort, we quantified a panel of serum cytokines prior to treatment initiation and evaluated their associations with PFS and OS to identify immune biomarkers that reflect host-tumor interactions and may complement more established tumor-based predictors of immunotherapy response. We present this article in accordance with the REMARK reporting checklist (available at https://tlcr.amegroups.com/article/view/10.21037/tlcr-2026-0446/rc).

Methods

Study design and patient cohort

This was a prospective, single-institution, observational cohort study conducted at the Veterans Affairs Ann Arbor Healthcare System (VAAAHS). The study was designed to identify serum biomarkers predictive of therapeutic response and toxicity in patients with metastatic NSCLC receiving ICIs. Enrollment occurred between November 2015 and June 2022. Eligible participants had histologically confirmed metastatic NSCLC and were initiating treatment with PD-1 or PD-L1 inhibitors, either as monotherapy or in combination with chemotherapy. A total of 96 patients with available baseline cytokine and chemokine data and complete clinical outcomes were included in this analysis (Table 1). Study size was limited by patients enrolled for the trial and specified selection criteria. Patients were followed for a median of 30 months (June 2022 through November 2024).

Ethics and regulatory compliance

The study was conducted in accordance with institutional policies, VA research regulations, and the Declaration of Helsinki and its subsequent amendments. The study was approved by the VA Ann Arbor Healthcare System Institutional Review Board (IRB #1597432), and all participants provided written informed consent prior to enrollment.

Cytokine and chemokine quantification

Baseline blood (10 mL) was collected for Veterans undergoing therapy with anti-PD1 or anti-PD-L1 based therapy (Table 2); samples were centrifuged and stored at −80 ℃ until batch analyzed at once for serum cytokines and chemokines. Samples did not undergo repeated freeze-thaw cycles and no additional procedures such as sample dilution were required. Baseline serum concentrations of 70 cytokines and chemokines were measured in duplicate using Luminex, a multiplex ELISA platform (Human Cytokine/Chemokine/Growth Factor Panel A Kit, Millipore Sigma, Cat# HCYTA-60K). Samples were run in technical duplicates and were processed by laboratory personnel blinded to clinical outcomes. Standards were run for each plate to avoid batch effects and were prepared as serial dilutions. A blank was also included. The study was blinded, where samples were prepared by one technician and analyzed by another technician. Cytokine concentrations were quantified based on the standard curve generated from the fluorescent signal from the serial dilution of standards. The panel included innate and adaptive immune mediators, EGF, Eotaxin, G-CSF, GM-CSF (undetectable), IFN-α2, IFN-β, IFN-γ, IFN-ω, IL-1α, IL-1β, IL-1RA, IL-2, IL-3, IL-4, IL-5, IL-6, IL-7, IL-8, IL-10, IL-12p40, IL-12p70, IL-13, IL-15, IL-17A, IL-17E, IL-17F, IL-18, IL-22, CXCL10 (IP-10), MCP-1, M-CSF, CXCL9, (MIG), CCL3 (MIP-1α), CCL4 (MIP-1β), PDGF-α, PDGF-β, RANTES, TNF-α, TNF-β, VEGF-α, CCL21 (6Ckine), CXCL13 (BCA-1), sFAS, sFASL, sCD137, APRIL, BAFF, IL-16, CCL27 (CTACK), CXCL5 (ENA-78), Eotaxin-2, Eotaxin-3, HMGB1, I-309, IL-23, IL-20, IL-28A, IL-31, IL-29, IL-33, IL-21, LIF, CCL8 (MCP-2), CCL15 (MIP-1δ), CCL20 (MIP-3α), CXCL11 (ITAC), SCF, SCF-1, CCL17 (TARC), TPO, TRAIL, TSLP, Perforin, Granzyme A, and Granzyme B.

Statistical analysis

All analyses were performed using R software, version 4.3.1. Cytokine concentrations were natural-log-transformed to reduce skewness and stabilize variance. Initially, separate Cox proportional hazards (PH) models were used for each cytokine to evaluate those associated with PFS or OS. PFS and OS for all patients, regardless of their enrollment year [2015–2022], were strictly calculated from each patient’s individual date of ICI treatment initiation (Time 0) to the date of disease progression, death, or last clinical contact. Each model was adjusted for age, sex, smoking history, disease stage (de novo vs. recurrent), histology, and treatment type (ICI monotherapy vs. chemoimmunotherapy). Hazard ratios (HRs) were expressed per doubling in cytokine concentration for interpretability. HRs were expressed per doubling in cytokine concentration for interpretability. To accurately derive this, the estimated coefficients from the models were scaled by a factor of ln[2] prior to exponentiation.

To control for multiple comparisons, false discovery rate (FDR) correction was applied using the Benjamini-Hochberg method. Cytokines remaining significant after FDR adjustment were incorporated into multivariable Cox models. To prevent model overfitting and maintain an adequate events-per-variable ratio, these final models included only the cytokines without further adjustment for the baseline clinical covariates. Model discrimination was assessed by time-dependent area under the curve (AUC) and Brier scores, estimated using 100 bootstrap cross-validation replicates.

Predictive accuracy was evaluated at 1-, 2-, and 3-year time points and at event-time percentiles (25^th, 50^th, 70^th). Significance was defined as a two-sided P<0.05 after FDR adjustment.

Exploratory latent class analysis (LCA)

To determine whether cytokine patterns defined distinct patient phenotypes, an exploratory LCA was performed (15). Cytokine concentrations were log-or power-law-transformed to optimize normality (as measured by the Shapiro-Wilk statistic), then standardized (centered and scaled by standard deviation). LCA was conducted using the tidySEM R package (v0.2.9) (16). Survival outcomes (PFS and OS) were compared between identified cytokine classes using Kaplan-Meier methods. The effect of cytokine-defined phenotype on PFS and OS was further evaluated using Cox regression, adjusting for age, sex, smoking history, disease stage, histology, and treatment type (ICI monotherapy vs. chemoimmunotherapy).

Results

Patient characteristics

A total of 98 patients with metastatic NSCLC were enrolled, of whom 96 (98%) had complete clinical data and adequate baseline serum samples available for cytokine and chemokine analysis. The cohort was predominantly male (96.9%) and Caucasian (83%), with a mean age of 66 years. Nearly all patients were current or former smokers (97.9%). Most patients received pembrolizumab (83.5%), including 14 who received pembrolizumab in combination with carboplatin and pemetrexed. Additional ICIs administered included durvalumab and nivolumab. Baseline demographic and clinical characteristics are summarized in Table 1.

Elevated IL-6 and IL-8 were associated with shorter PFS and OS

Higher baseline IL-6 and IL-8 concentrations were significantly associated with shorter PFS and OS (Table 3). Each doubling of IL-6 was associated with a 29% higher hazard of progression [HR =1.29; 95% confidence interval (CI): 1.09–1.53; P=0.002; FDR q=0.046] and a 34% higher hazard of death (HR =1.34; 95% CI: 1.13–1.60; P<0.001; FDR q=0.023). Similarly, each doubling of IL-8 corresponded to a 39% higher hazard of progression (HR =1.39; 95% CI: 1.18–1.62; P<0.001; FDR q=0.004) and a 36% higher hazard of death (HR =1.36; 95% CI: 1.16–1.59; P<0.001; FDR q=0.009). These findings demonstrate that elevated proinflammatory cytokines, particularly IL-6 and IL-8, are strongly associated with adverse outcomes in ICI-treated NSCLC (Tables 4,5).

Elevated IFN-ω and perforin are associated with longer PFS and OS

In contrast, higher baseline IFN-ω and perforin concentrations were associated with significantly improved outcomes. Each doubling of IFN-ω was associated with a 13% lower hazard of progression (HR =0.87; 95% CI: 0.80–0.95; P=0.002; FDR q=0.037) and a 13% lower hazard of death (HR =0.87; 95% CI: 0.80–0.95; P=0.001; FDR q=0.023). Perforin showed an even stronger protective association: each doubling reduced the risk of progression by 49% (HR =0.51; 95% CI: 0.35–0.73; P<0.001; FDR q=0.012) and the risk of death by 50% (HR =0.50; 95% CI: 0.33–0.76; P<0.001; FDR q=0.023). Stem cell factor-1 (SCF-1) was also associated with improved PFS (HR =0.51; 95% CI: 0.33–0.80; P=0.003; FDR q=0.046) but not with OS (Tables 6,7).

Multivariable cytokine models

When cytokines significantly associated with survival were analyzed simultaneously in multivariable Cox models, IL-6 and IL-8 remained independently associated with shorter PFS and OS, whereas IFN-ω and perforin retained protective associations (Tables 3,8). SCF-1 demonstrated a trend toward reduced progression risk but did not reach statistical significance. The multivariable cytokine model demonstrated strong prognostic discrimination for PFS, with time-dependent AUCs of 82.0% at 1 year, 76.5% at 2 years, and 78.3% at 3 years (Table 9). For OS, corresponding AUCs were 77.6%, 75.6%, and 68.6% at 1, 2, and 3 years, respectively (Table 10; Figures 1,2).

Figure 1 Forest plot of cytokines for progression and death. Left (PFS) and right (OS) panels display adjusted HRs per doubling of cytokine concentration from a multivariable Cox model. The PFS model included IL-6, IL-8, IFN-ω, SCF-1, and perforin; the OS model included IL-6, IL-8, IFN-ω, and perforin. Points show HR per doubling of cytokine level; bars are 95% CIs; dashed vertical line indicates HR =1. CI, confidence interval; HR, hazard ratio; IFN-ω, interferon-omega; IL-6, interleukin-6; IL-8, interleukin-8; OS, overall survival; PFS, progression-free survival; SCF-1, stem cell factor-1.

Figure 2 Time-dependent predictive performance of cytokine panel for progression and death. Time-dependent AUC and Brier scores for the multivariable cytokine model predicting PFS. The panel, including IL-6, IL-8, perforin, SCF-1, and IFN-ω, demonstrated strong discrimination with AUC ≥0.80 at 1 year and sustained predictive accuracy through 3 years. AUC, area under the curve; IFN-ω, interferon-omega; IL-6, interleukin-6; IL-8, interleukin-8; PFS, progression-free survival; SCF-1, stem cell factor-1.

LCA

Exploratory LCA was performed using five selected cytokines (IL-6, IL-8, perforin, IFN-ω, and TRAIL) chosen based on distributional normalization and minimal cross-correlation. Two distinct patient phenotypes were identified: Phenotype 1, enriched for IL-6 and IL-8, and Phenotype 2, enriched for perforin, IFN-ω, and TRAIL. LCA assigned 24 patients to Phenotype 1 and 72 patients to Phenotype 2. Baseline demographic and clinicopathologic characteristics were similar between phenotypes, with no statistically significant differences observed across evaluated variables (Table 11). Patients in the IL-6/IL-8-enriched phenotype had significantly shorter PFS and OS compared with those in the perforin/IFN-ω/TRAIL-enriched phenotype (Figure 3, log-rank P<0.001 for both). To derive the standard value for Figure 3, the raw values were power-law transformed to maximize normality by the Shapiro-Wilk test, centered the mean on zero, and divided by the standard deviation. In multivariable analyses adjusting for clinical covariates, phenotype remained significantly associated with PFS (HR =2.42; 95% CI: 1.43–3.99) and OS (HR =2.44; 95% CI: 1.69–5.00). These findings confirm the existence of biologically distinct, cytokine-defined immune classes with prognostic relevance in ICI-treated NSCLC.

Figure 3 LCA. Exploratory LCA identified 2 phenotypes defined by high IL-6 and IL-8 and low perforin, IFN-ω and TRAIL (24 patients), and conversely low IL-6 and IL-8 and high perforin, IFN-ω and TRAIL (72 patients, A). Survival analysis revealed that patients with phenotype 1 had significantly shorter progression-free survival (B, Mantel-Cox P<0.001) and OS (C, Mantel-Cox P<0.001). IFN-ω, interferon-omega; IL-6, interleukin-6; IL-8, interleukin-8; LCA, latent class analysis; OS, overall survival; TRAIL, tumor necrosis factor-related apoptosis-inducing ligand.

Discussion

Key findings

This study demonstrates that baseline cytokine and chemokine profiles carry strong prognostic significance in patients with metastatic NSCLC treated with ICIs. Elevated IL-6 and IL-8 were consistently associated with worse outcomes, whereas higher levels of IFN-ω and perforin were linked to improved survival. These findings underscore the potential dual role of circulating immune mediators: while pro-inflammatory cytokines such as IL-6 and IL-8 promote tumor progression and immune resistance, immune-activating factors like IFN-ω and perforin may reflect an intact antitumor response.

Comparison with similar research

Our results reinforce prior work implicating IL-6 and IL-8 as markers of poor prognosis in NSCLC. IL-6 promotes tumor progression through JAK-STAT3 signaling, angiogenesis, and induction of an immunosuppressive tumor microenvironment (17). More recently, IL-6 has been implicated in checkpoint inhibitor resistance via cancer-induced neuroimmune signaling (18). IL-8, a potent neutrophil chemoattractant, enhances tumor cell proliferation and angiogenesis and recruits myeloid-derived suppressor cells, thereby dampening ICI efficacy (19).

Explanation of findings

Importantly, in our multivariable models, IL-6 and IL-8 remained independent predictors of outcome after adjusting for clinical covariates, suggesting their potential utility as clinically actionable biomarkers. Conversely, IFN-ω and perforin emerged as protective biomarkers, consistent with their roles in cytotoxic immune activity. Type I interferons, including IFN-ω, enhance antigen presentation and activate cytotoxic T lymphocytes, thereby potentiating antitumor immunity. Our data show that higher baseline IFN-ω levels were associated with improved survival, supporting the hypothesis that type I interferon signaling augments ICI efficacy. These findings align with prior reports linking type I interferon gene signatures to favorable ICI responses (20,21). Although less well characterized than IFN-α or IFN-β, IFN-ω exerts potent immunostimulatory effects by inducing interferon-stimulated genes, promoting antigen presentation, and enhancing cytotoxic T-cell priming. This study provides the first clinical evidence that IFN-ω may serve as a novel biomarker of immunotherapy responsiveness.

Perforin, a cytolytic granule protein secreted by activated CD8⁺ T cells and natural killer cells, is essential to the effector phase of tumor cell killing. Its association with improved PFS and OS highlights the importance of an intact cytotoxic immune response in determining benefit from ICIs. Interestingly, SCF-1 also correlated with improved PFS, though its mechanistic role in NSCLC immune regulation remains poorly defined and warrants further investigation.

Implications and action needed

The LCA adds an innovative layer of insight by identifying two distinct immunologic phenotypes: one enriched for IL-6 and IL-8 with inferior outcomes, and another enriched for IFN-ω, perforin, and TRAIL with superior survival. This data-driven stratification suggests that baseline cytokine patterns can define biologically distinct patient subgroups, potentially informing therapeutic tailoring. Patients with an IL-6/IL-8-dominant “pro-inflammatory” profile may represent a distinct biological subgroup with poorer outcomes following ICI therapy, whereas cytotoxic cytokine-enriched “immune-active” profiles were associated with more favorable clinical outcomes.

Collectively, these findings support the prognostic value of circulating cytokine and chemokine profiling in NSCLC and provide a rationale for incorporating cytokine signatures into future biomarker-driven clinical trials. Prospective validation in larger, multi-institutional cohorts—and integration with tumor- and host-derived biomarkers such as PD-L1, TMB, radiomic features, and the gut microbiome—will be essential to confirm their predictive utility and translate cytokine-based models into clinically actionable tools.

Strengths and limitations

This study has several notable strengths. First, we utilized high-dimensional multiplex cytokine profiling, allowing simultaneous evaluation of a broad range of immune and inflammatory mediators rather than reliance on single-cytokine analyses, thereby providing a more comprehensive assessment of host immune state. Additionally, the use of LCA represents an innovative, data-driven approach to identifying biologically distinct cytokine-defined immune phenotypes with prognostic relevance, extending beyond traditional biomarker analysis alone. The observed cytokine associations are consistent with studies in the literature, aligning with known roles of IL-6/IL-8-mediated inflammation, supporting the translational relevance of these findings.

This study also has several limitations. First, the sample size was modest and drawn from a single-institution Veteran population that was predominantly male and Caucasian, which may limit generalizability. Second, it is important to note that while our samples did not undergo repeated freeze-thaw cycles, it is possible that disparate storage time could affect the results. Third, cytokine measurements were restricted to baseline; longitudinal profiling during treatment could yield additional insights into response and resistance mechanisms. Fourth, although our models demonstrated strong prognostic performance, our results are exploratory given need for future external validation in larger, more diverse cohorts, particularly given that nearly all patients were tobacco users. Additionally, several potentially relevant cofounders were not fully captured in this cohort. PD-L1 expression data and body mass index (BMI), both of which have been associated with immunotherapy outcomes and immune signaling (22,23), were not consistently available. As a result, any residual confounding cannot be excluded and may have influenced observed cytokine-survival associations. Finally, future work should evaluate integrated models combining cytokine signatures with genomic, transcriptomic, and radiomic biomarkers to develop comprehensive, multi-dimensional predictors of ICI efficacy.

Conclusions

In this prospective study of metastatic NSCLC treated with ICIs, baseline cytokine and chemokine profiles were strongly prognostic for clinical outcomes. Elevated IL-6 and IL-8 were independently associated with higher hazards of progression and death, while higher IFN-ω and perforin levels predicted improved survival, reflecting distinct pro-tumorigenic vs. cytotoxic immune states. LCA further revealed immunologic subgroups with divergent trajectories, suggesting that cytokine signatures capture biologically meaningful heterogeneity not reflected by PD-L1 or TMB alone.

These findings highlight the potential of circulating cytokines as accessible, minimally invasive biomarkers that complement existing tumor-based assays to refine patient risk stratification and therapeutic decision-making. Validation in larger, multi-institutional cohorts and integration with molecular and imaging biomarkers represent essential next steps. Ultimately, our data suggests that cytokine-guided biomarker models may have value in risk stratification and rational development of combinatorial strategies targeting IL-6/IL-8-driven inflammatory pathways.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the REMARK reporting checklist. Available at https://tlcr.amegroups.com/article/view/10.21037/tlcr-2026-0446/rc

Data Sharing Statement: Available at https://tlcr.amegroups.com/article/view/10.21037/tlcr-2026-0446/dss

Peer Review File: Available at https://tlcr.amegroups.com/article/view/10.21037/tlcr-2026-0446/prf

Funding: This work was supported by the Department of Veterans Affairs (VA) Merit Award (No. 1I01 CX001560-01A1).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://tlcr.amegroups.com/article/view/10.21037/tlcr-2026-0446/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with institutional policies, VA research regulations, and the Declaration of Helsinki and its subsequent amendments. The study was approved by the VA Ann Arbor Healthcare System Institutional Review Board (IRB #1597432), and all participants provided written informed consent prior to enrollment.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Rother C, John T, Wong A. Biomarkers for immunotherapy resistance in non-small cell lung cancer. Front Oncol 2024;14:1489977. [Crossref] [PubMed]
2. Wang X, Lamberti G, Di Federico A, et al. Tumor mutational burden for the prediction of PD-(L)1 blockade efficacy in cancer: challenges and opportunities. Ann Oncol 2024;35:508-22. [Crossref] [PubMed]
3. Wang M, Zhai X, Li J, et al. The Role of Cytokines in Predicting the Response and Adverse Events Related to Immune Checkpoint Inhibitors. Front Immunol 2021;12:670391. [Crossref] [PubMed]
4. Sanmamed MF, Perez-Gracia JL, Schalper KA, et al. Changes in serum interleukin-8 (IL-8) levels reflect and predict response to anti-PD-1 treatment in melanoma and non-small-cell lung cancer patients. Ann Oncol 2017;28:1988-95. [Crossref] [PubMed]
5. Keegan A, Ricciuti B, Garden P, et al. Plasma IL-6 changes correlate to PD-1 inhibitor responses in NSCLC. J Immunother Cancer 2020;8:e000678. [Crossref] [PubMed]
6. Ayers M, Lunceford J, Nebozhyn M, et al. IFN-γ-related mRNA profile predicts clinical response to PD-1 blockade. J Clin Invest 2017;127:2930-40. [Crossref] [PubMed]
7. Rooney MS, Shukla SA, Wu CJ, et al. Molecular and genetic properties of tumors associated with local immune cytolytic activity. Cell 2015;160:48-61. [Crossref] [PubMed]
8. Liu Q, Shaibu Z, Xu A, et al. Predictive value of serum cytokines in patients with non-small-cell lung cancer receiving anti-PD-1 blockade therapy: a meta-analysis. Clin Exp Med 2025;25:59. [Crossref] [PubMed]
9. Kang DH, Park CK, Chung C, et al. Baseline Serum Interleukin-6 Levels Predict the Response of Patients with Advanced Non-small Cell Lung Cancer to PD-1/PD-L1 Inhibitors. Immune Netw 2020;20:e27. [Crossref] [PubMed]
10. Laino AS, Woods D, Vassallo M, et al. Serum interleukin-6 and C-reactive protein are associated with survival in melanoma patients receiving immune checkpoint inhibition. J Immunother Cancer 2020;8:e000842. [Crossref] [PubMed]
11. Mao XC, Yang CC, Yang YF, et al. Peripheral cytokine levels as novel predictors of survival in cancer patients treated with immune checkpoint inhibitors: A systematic review and meta-analysis. Front Immunol 2022;13:884592. [Crossref] [PubMed]
12. Qu Z, Sun F, Zhou J, et al. Interleukin-6 Prevents the Initiation but Enhances the Progression of Lung Cancer. Cancer Res 2015;75:3209-15. [Crossref] [PubMed]
13. Ota T, Fukui T, Nakahara Y, et al. Serum immune modulators during the first cycle of anti-PD-1 antibody therapy in non-small cell lung cancer: Perforin as a biomarker. Thorac Cancer 2020;11:3223-33. [Crossref] [PubMed]
14. Yu R, Zhu B, Chen D. Type I interferon-mediated tumor immunity and its role in immunotherapy. Cell Mol Life Sci 2022;79:191. [Crossref] [PubMed]
15. Sinha P, Calfee CS, Delucchi KL. Practitioner's Guide to Latent Class Analysis: Methodological Considerations and Common Pitfalls. Crit Care Med 2021;49:e63-79. [Crossref] [PubMed]
16. Van Lissa CJ, Góngora VM, Anadria DJ. Recommended practices in latent class analysis using the open-source R-package tidySEM. Struct Equ Model Multidiscip J 2024;31:526-34.
17. Thuya WL, Cao Y, Ho PC, et al. Insights into IL-6/JAK/STAT3 signaling in the tumor microenvironment: Implications for cancer therapy. Cytokine Growth Factor Rev 2025;85:26-42. [Crossref] [PubMed]
18. Baruch EN, Gleber-Netto FO, Nagarajan P, et al. Cancer-induced nerve injury promotes resistance to anti-PD-1 therapy. Nature 2025;646:462-73. [Crossref] [PubMed]
19. Teijeira A, Garasa S, Ochoa MC, et al. IL8, Neutrophils, and NETs in a Collusion against Cancer Immunity and Immunotherapy. Clin Cancer Res 2021;27:2383-93. [Crossref] [PubMed]
20. Fernández-García F, Fernández-Rodríguez A, Fustero-Torre C, et al. Type I interferon signaling pathway enhances immune-checkpoint inhibition in KRAS mutant lung tumors. Proc Natl Acad Sci U S A 2024;121:e2402913121. [Crossref] [PubMed]
21. Lim J, La J, Kim HC, et al. Type I interferon signaling regulates myeloid and T cell crosstalk in the glioblastoma tumor microenvironment. iScience 2024;27:110810. [Crossref] [PubMed]
22. Wang Z, Aguilar EG, Luna JI, et al. Paradoxical effects of obesity on T cell function during tumor progression and PD-1 checkpoint blockade. Nat Med 2019;25:141-51. [Crossref] [PubMed]
23. Cristescu R, Mogg R, Ayers M, et al. Pan-tumor genomic biomarkers for PD-1 checkpoint blockade-based immunotherapy. Science 2018;362:eaar3593. [Crossref] [PubMed]