Immunotherapy has revolutionized the treatment in head and neck squamous cell carcinoma (HNSCC). Immune checkpoints inhibitors (ICI) that target the programmed death-1 (PD-1) axis have shown unprecedented rates of durable responses in relapsed and metastatic (R/M) HNSCC. Indeed, pembrolizumab monotherapy or in combination with platinum-based chemotherapy has been approved in the first line setting for R/M HNSCC and nivolumab has received approval in the second line platinum-refractory setting. However, a significant number of treated patients do not respond, and the potential for serious side effects remains. There is a growing imperative to identify biomarkers that can enhance the selection of patients likely to respond optimally to therapy, provide deeper insights into drug mechanisms of action, and facilitate the customization of therapy regimens.Currently, only combined positivity score (CPS) for PD-L1, a score based on the ratio between tumour cells and immune cells expressing PD-L1 to define tumour PD-L1 positivity, has shown a positive correlation with response to pembrolizumab and survival in the phase III KEYNOTE-048 study. In KEYNOTE-040, the correlation with clinical outcome was also strongly positive when using PD-L1 expression in tumour cells only (TPS 50%), congruent with the experience in non-small cell lung cancer in KEYNOTE-010. In contrast, no correlation was found in the nivolumab CHECKMATE-141 study, where PD-L1 expression was exclusively determined in tumour cells. Furthermore, PD-L1 negative patients may still benefit from ICI in combination with chemotherapy. Crucially, the regulation of PD-L1 expression involves various signalling pathways, including MAPK, PI3K, and Akt/PKB, which are frequently altered in HNSCC. Due to these molecular crosstalks, PD-L1 emerges as a dynamic biomarker, exhibiting temporal variations and spatial heterogeneity. Its expression may undergo changes from initial diagnosis to recurrence or progression and may vary between primary and coexisting metastatic lesions. Indeed, conflicting results regarding the intra-tumoural heterogeneity of PD-L1 expression in HNSCC are evident in published reports. Therefore, additional factors beyond PD-L1 expression might also contribute to treatment response. On this regard, some preliminary studies have shown that tumour mutational burden (TMB), T-cell-inflamed gene expression profile and human papillomavirus (HPV) status may correlate with response to ICI. In addition, somatic mutations with frameshift events in tumour suppressors (i.e. NOTCH1, SMARCA4) appear significantly enriched in among HPV-negative responders. Type I interferons (IFNs) play a crucial role in the mechanism through which the innate immune sensing of tumours leads to recruitment of cytotoxic T cells, a pivotal step in establishing an inflamed tumour microenvironment. In the context of HNSCC, the relationship between IFNs and response to ICI was explored in the KEYNOTE-O12 trial, where the results revealed a statistically significant association between the pre-treatment IFN-γ gene signature (including IDO1, CXCL10, CXCL9, HLA-DRA, STAT1, IFN-γ gene expression) and best overall response and progressionfree survival, suggesting its potential as a biomarker for patient exclusion from immunotherapy, given its high negative predictive value.Although tumour tissue-based biomarkers are extensively employed for identifying patients with enhanced responsiveness to ICI, several challenges persist in clinical practice such as its invasive nature (constrained by tumour accessibility, patient's overall condition and potential procedure-related complications), spatial heterogeneity and low tumour content. Furthermore, the dynamic nature of cancer immunity during immunotherapy makes longitudinal monitoring through repeated biopsies impractical in the clinical setting. Consequently, clinicians typically base decisions on a pre-treatment single-timepoint tumour biopsy, rather than conducting repeated biopsies to track the evolving immunological profiles. The emergence of high throughput multiplexed analytical technologies has made peripheral blood a viable source for more comprehensive immune profiling. Peripheral blood sampling, being readily available, minimally invasive, and repeatable, offers a means to address the aforementioned limitations of tissue-based biomarkers. On this regard, circulating biomarkers such as serum proteins and cytokines, circulating immune and tumour cells, and T-cell receptor repertoire dynamics, are under exploration. Finally, host-related markers have been gradually explored. These include general characteristics (i.e. sex, age and performance status), host germline genetics and intestinal commensal microbiota. Accumulating evidence suggests that the intestinal microbiota can influence host anti-cancer immune responses and impact the efficacy of anticancer therapies, including immunotherapy. The role of the microbiota in predicting the response to ICI in HNSCC is yet to be fully understood. A single sub-study from CHECKMATE-141 that explored the oral microbiota measured in saliva as a potential predictive biomarker in patients with R/M HNSCC treated with nivolumab showed no significant correlation with treatment efficacy or survival, but it had limitations such as non-uniform sample collection, small number of responses for correlation, and notably and omission of the analysis of the intestinal microbiota. Considering the immunomodulatory effects of the intestinal microbiota and the emerging evidence of the oral microbiota influencing HNSCC tumorigenesis and progression, there is a need to study their role as predictive biomarkers for the response to ICI in HNSCC.The pursuit of reliable biomarkers is constrained by our incomplete understanding of how immunotherapies modify the intricate immune response to cancer, along with the impact of immuno-editing on a dynamic and inducible tumour microenvironment and immune milieu. Moreover, the limited extension of candidate assays into large, prospective studies and the lack of standardization in measurement and interpretation curtail their validity.