Introduction
Oral squamous cell carcinoma (OSCC) is the most common malignancy of the oral cavity and accounts for approximately 90% of all oral cancers (1–3). It represents a major subgroup of head and neck squamous cell carcinomas (HNSCCs), which are among the sixth most common cancers worldwide, with over 500,000 new cases annually (2). Despite advances in surgery, chemotherapy, and radiotherapy, the prognosis of patients with recurrent or metastatic OSCC (R/M-OSCC) remains poor.
Since the 1980s, platinum-based chemotherapy regimens have served as the standard treatment for advanced head and neck cancers. The EXTREME regimen, which combines 5-fluorouracil, cisplatin, and cetuximab, became a standard of care in 2012 (4). More recently, immune checkpoint inhibitors (ICIs), such as nivolumab and pembrolizumab, targeting the programmed death-1 (PD-1)/programmed death-ligand 1 (PD-L1) pathway, have expanded the therapeutic landscape and shown promising efficacy in R/M-HNSCC, including OSCC (5,6). These agents are now widely used in clinical practice in Japan and globally. However, ICIs are not universally effective, and their clinical benefit is limited to a subset of patients (7). Therefore, there is a pressing need to identify reliable biomarkers to predict which patients are most likely to respond to ICI therapy. While previous studies have proposed PD-L1 expression, tumor mutational burden, and tumor-infiltrating lymphocytes (TILs) as candidate predictive biomarkers, none have demonstrated sufficient sensitivity or specificity for routine clinical use (8–10).
In recent years, inflammation- and nutrition-based biomarkers have emerged as potential prognostic indicators in various malignancies, including head and neck cancers (11–25). These include the neutrophil-to-lymphocyte ratio (NLR) (13–16,23,25–30), platelet-to-lymphocyte ratio (PLR) (14–18,23,24), lymphocyte-to-monocyte ratio (LMR) (25), C-reactive protein-to-albumin ratio (CAR) (12,20), prognostic nutritional index (PNI) (31,32), and modified Glasgow Prognostic Score (mGPS) (22,24). These markers are simple to calculate using routinely available laboratory data and reflect the complex interplay between systemic inflammation, nutritional status, and cancer progression. Although several reports have explored the prognostic significance of these markers in HNSCC, limited data exist regarding their role in predicting the response and outcomes of ICI treatment specifically in patients with OSCC. Moreover, it remains unclear whether the timing of biomarker evaluation, before or after ICI initiation, affects their prognostic utility.
Therefore, the aim of this retrospective study was to evaluate the clinical relevance of inflammation- and nutrition-based biomarkers as predictors of treatment response and prognosis in patients with R/M-OSCC receiving ICI therapy. We also investigated whether biomarker levels assessed after ICI administration offer superior prognostic value compared to those assessed prior to treatment.
Discussion
In this study, pre-treatment IBPS values did not show significant associations with OS or PFS. However, post-ICI treatment values for NLR, PNI, CAR, and mGPS were significantly correlated with OS. Similarly, post-treatment values for NLR, PNI, and CAR were significantly associated with PFS. While no significant association was observed between pre-treatment IBPS values and the DCR, post-treatment values of NLR, LMR, PNI, PLR, and CAR were significantly correlated with DCR. Survival analysis using Kaplan–Meier curves, with cutoff values derived from ROC analysis, revealed that high NLR, low LMR, high PLR, high CAR, and low lymphocyte counts prior to ICI administration were associated with shorter OS. Furthermore, high NLR, low LMR, and low lymphocyte counts following ICI administration were significantly associated with shorter OS.
NLR is a marker of systemic inflammation and has been reported as a prognostic factor in various cancers, including R/M-OSCC (26–30). In our study, post-ICI NLR demonstrated the highest predictive accuracy (AUC=0.7696), with patients exhibiting NLR ≥7.17 having significantly shorter OS than those with NLR <7.17. An elevated NLR reflects an increase in neutrophils and a decrease in lymphocytes, suggesting tumor-associated inflammation and immune suppression (35). Several studies have reported that an increase in TILs is associated with a favorable prognosis (29,30). The mechanism underlying the association between elevated NLR and poor prognosis may be attributed to the inflammatory response, in which increased neutrophil production promotes tumor proliferation, invasion, and angiogenesis by inducing the secretion of chemokines and cytokines, further enhancing tumor-associated inflammation and suppressing anti-tumor immunity.
CAR is a marker that reflects systemic inflammation and nutritional status. In patients with a high CAR, elevated inflammatory cytokines stimulate CRP production while inhibiting albumin synthesis and accelerating its degradation, signaling the progression of cachexia (36–38). In head and neck squamous cell carcinoma, high interleukin-6 (IL-6) expression has been reported, and IL-6 signaling promotes immunosuppression and tumor progression (39,40). Local and systemic increases in IL-6 correlate with elevated CRP concentrations across various cancers (41,42). In our study, a high CAR was associated with a poor prognosis, highlighting its potential as an important prognostic marker for patients with squamous cell carcinoma undergoing chemoradiotherapy or cytotoxic chemotherapy.
Similarly, mGPS, calculated based on serum albumin and CRP levels (43), has been recognized as a prognostic factor for patients with recurrent or metastatic head and neck cancer (44–48). Consistent with these findings, our study also demonstrated a significant association between post-ICI mGPS values and OS.
PLR, another hematological inflammatory marker, is calculated from platelet and lymphocyte counts. Platelets, similar to neutrophils, play a key role in inflammation and are often elevated in patients with chronic inflammation associated with solid tumors (49,50). Elevated platelet counts have been linked to cancer progression, making PLR a promising predictive marker for tumor progression (31,32). Several studies have suggested that preoperative PLR values are valuable for prognosis prediction in gastrointestinal cancers, and more recent research has highlighted its potential for predicting the efficacy of systemic chemotherapy (14).
PNI, calculated using albumin levels and lymphocyte counts, reflects nutritional and immunological statuses (51). Previous studies have demonstrated that pre-ICI PNI is associated with the efficacy of ICIs in advanced or recurrent non-small cell lung cancer (52). Given the strong relationship between immune function and nutritional status, low PNI values are typically associated with poor prognosis. Our study further corroborates this, demonstrating that low PNI values are associated with reduced ICI efficacy and worse prognosis.
Notably, in our study, the AUC of nearly all hematological markers improved after treatment compared with pre-treatment values. This suggests that post-treatment values may serve as more reliable prognostic indicators in patients receiving ICI therapy. Matsuo et al reported that pre-treatment IBPSs were insufficient for predicting prognosis in patients with R/M-head and neck squamous cell carcinoma, while post-nivolumab GPS, NLR, CAR, and eosinophil counts were significantly associated with survival outcomes (48). Similarly, Tachinami et al emphasized the importance of post-treatment IBPS, including NLR, CAR, and PLR, in predicting prognosis in R/M-OSCC (53). Our findings are consistent with those of these studies, suggesting that post-treatment data could be instrumental in predicting treatment response, enabling the early identification of patients with poor responses to ICI therapy and allowing for timely transition to alternative treatment strategies. In this study, the AUC values of biomarkers such as NLR, PNI, and CAR were below 0.8. While an AUC above 0.8 is considered to indicate high predictive accuracy, biomarkers with AUC values between 0.6 and 0.8 can still be clinically useful (54,55). Especially in patients with R/M-OSCC undergoing ICI therapy, the tumor immune microenvironment is complex, and it is challenging to predict prognosis with a single biomarker alone. In our analysis, even though the AUC values were moderate, NLR and PNI were found to be independent prognostic factors in multivariable analysis. Therefore, these markers may still have clinical value in supporting treatment decisions. Further improvements in predictive performance may be achieved by developing composite models that integrate multiple clinical and biological factors. Recent studies have identified several factors influencing the efficacy of ICIs, including irAEs, BMI, history of radiotherapy, ECOG PS, and CPS for PD-L1 expression (38,48,56–60). In our study, sex and BMI were significant predictors of OS, whereas sex, ICI type, presence of irAEs, and BMI were significant predictors of PFS. Additionally, ECOG PS, CPS values, BMI, and radiotherapy history were significantly associated with DCR.
In addition to hematological biomarkers such as IBPS, various other factors have been proposed as potential predictors of response to ICIs. Given the accessibility of peripheral blood samples, numerous studies have analyzed circulating cytokine levels, including tumor necrosis factor-alpha, interferon-γ (IFN-γ), IL-6, IL-8, and transforming growth factor-β (TGF-β) (61–63). Elevated IFN-γ levels have been associated with improved ICI efficacy and toxicity, while higher baseline levels of IL-8, IL-6, and TGF-β have been identified as negative predictive markers. However, the extent to which these circulating factors accurately reflect the tumor microenvironment remains debatable. Additionally, studies have indicated that a high neoantigen burden correlates with better treatment outcomes (64), whereas genetic mutations (e.g., epidermal growth factor receptor, KRAS, PTEN, TP53) and dysregulation of the Wnt signaling pathway may impact ICI response (65–69). Moreover, the use of corticosteroids, antibiotics, or proton pump inhibitors influences ICI efficacy, potentially through alterations in gut microbiota composition (70–72). Future research should aim to integrate these molecular, clinical, and therapeutic factors to optimize patient selection and enhance treatment strategies for ICI therapy.
One major limitation of this study is the small sample size, particularly after stratification into subgroups. This limitation is inherent to the retrospective design and the rarity of patients with R/M-OSCC treated with ICIs at a single institution. Although the limited cohort size reduced statistical power, we were still able to identify consistent trends in the performance of prognostic and predictive biomarkers, especially for NLR and PNI. These results aligned with those of previous reports (31,32,48,53) and suggest that inflammation- and nutrition-based biomarkers may provide meaningful clinical insights, even in smaller populations. In addition, potential selection biases related to treatment history, clinical stage, and histological type were not fully controlled. Thus, further validation in prospective, multi-center studies with larger cohorts is essential to confirm our findings and develop robust predictive models. This study primarily aimed to identify pre-treatment biomarkers predictive of ICI efficacy in R/M-OSCC. While we also evaluated post-treatment biomarkers, their prognostic value seemed limited, likely due to their reflection of early treatment dynamics rather than baseline tumor biology. Nevertheless, post-treatment biomarkers assessed 4–6 weeks after ICI initiation showed potential for early prediction of treatment efficacy, which may help identify non-responders and guide timely treatment modifications. As our study did not include a control group receiving alternative treatments such as chemotherapy or targeted agents, the generalizability of our findings across different treatment modalities remains uncertain. Future prospective, multi-arm studies are warranted to evaluate the prognostic consistency of these biomarkers and to determine whether post-treatment markers retain predictive value independent of treatment type.
In conclusion, our study demonstrated that post-ICI values of NLR, PNI, CAR, and mGPS were significantly associated with survival outcomes in patients with R/M-OSCC. While pre-treatment IBPS values did not demonstrate significant associations with DCR, post-ICI values of NLR, LMR, PNI, PLR, and CAR exhibited significant correlations. These findings suggest that hematological biomarkers, including IBPS, may serve as valuable predictors of ICI treatment efficacy. Early prediction of ICI response and timely adjustments to treatment may lead to improved prognosis and the early identification of patients who could benefit from sequential treatment.