Introduction
Head and neck cancer (HNC), ranking as the sixth most common cancer globally, accounts for approximately 30,000 fatalities each year [
1]. The predominant pathological subtype of HNC is squamous cell carcinoma [
2]. In advanced cases of head and neck squamous cell carcinoma (HNSCC), the primary causes of mortality are local recurrence, cervical lymph node metastases, and resistance to conventional chemotherapy, which often leads to treatment failure. In HNSCC, the tumor microenvironment (TME) is composed of altered tumor cells, immune cells, and stromal cellular components [
3]. Extensive research on TME has highlighted the pivotal role of tumor-infiltrating immune cells in tumor progression, recurrence, metastasis, and response to immunotherapy treatments [
4].
The role of microbiota as a key regulator in immune cell activation, inflammation, and cancer progression has been increasingly recognized. Specifically, microbiota influences these biological processes through mechanisms involving nuclear factor kappa B (NF-κB), type I interferon, and inflammasome activation [
5]. Moreover, the local immune system’s interaction with gut microbiota plays a crucial role in modulating immune responses, tissue damage, and the development of cancer, as evidenced by several studies [
6]. The emerging role of microbes in cancer research is increasingly becoming a focal point, offering profound insights into cancer development [
7,
8]. In addition to their intrinsic role, microbes are also being explored as a potential tool for adjunct diagnosis in cancer research [
9,
10]. Studies have shown that patients with higher loads of
Fusobacterium nucleatum (
F. nucleatum) DNA in cancerous tissues tend to have shorter survival durations, highlighting its potential as a biomarker for prognosis [
11]. The newly released Cancer Microbiome Atlas (TCMA) encompasses curated microbial profiles from a comprehensive collection of 3,689 samples across 1,772 patients, spanning five The Cancer Genome Atlas (TCGA) programs and 21 anatomical locations [
12]. This atlas has been actively utilized in research on various cancers, including gastric, colon, and HNSCC. Its application facilitates multi-omics studies, enabling systematic analyses of microbe-host interactions [
13].
To elucidate the impact of various factors on the formation and maintenance of the TME, as well as on clinical prognosis in HNSCC, we analyzed TME infiltration patterns using multi-omics data from the TCGA HNSCC cohort. This included correlating immune status with genetic and intratumor immune-related microbiome characteristics. Our findings provide valuable insights into the immunological and microbial landscape of HNSCC, and their impacts on patients’ prognosis. These findings have important implications for enhancing treatment strategies and improving patient outcomes.
Discussion
Recent studies have highlighted the pivotal role of tumor-infiltrating immune cells within the TME in influencing tumor progression and clinical prognosis. These cells are increasingly recognized as valuable targets for therapy [
21]. Consequently, strategies aimed at remodeling the tumor-immune microenvironment are emerging as promising approaches to augment the anti-tumor immune response [
22]. However, the characteristics of immune phenotyping and the underlying mechanisms remain to be elucidated.
In this study, we integrated the immune cell landscape of HNSCC and categorized patients into three groups based on their immune characteristics. The ICI-1 was characterized by a particularly high stromal fraction, especially in terms of fibroblast proportion. The ICI-2, with the lowest immune scores and the poorest prognosis, was often considered to represent ‘cold tumors.’ In contrast, the ICI-3 had the highest immune scores and the best prognosis, typically referred to as ‘hot tumors.’ We observed that within the TME, the ICI-2 subtype, characterized by lower proportions of CD8 T cells, memory B cells, Tfh cells, activated CD4 memory T cells, Treg cells, and naive B cells, exhibited a higher hypoxia score. Additionally, this subtype showed a relatively higher proportion of exhausted T cells. These findings suggest that intratumoral hypoxia may further exacerbate immune suppression. Recent studies indicated that the impact of hypoxia on the tumor immune microenvironment primarily affects the function and distribution of immune cells [
23]. Hypoxic conditions, through the activation of factors like HIF-1, can alter the activity of immune cells such as T cells and macrophages, diminishing their anti-tumor effects in the TME [
24,
25]. Additionally, hypoxia may induce phenotypic changes in tumor-infiltrating immune cells, for instance, promoting the accumulation of immunosuppressive cells like regulatory T cells (Tregs) and myeloid-derived suppressor cells (MDSCs), thereby inhibiting the immune system’s attack on the tumor [
26,
27]. These alterations collectively facilitate tumor immune evasion and can adversely affect the efficacy of cancer treatments.
A mounting amount of research suggests that the microbiome is essential for modifying immune responses to cancer treatment [
28]. Within the DEGs between ICI-2 and ICI-3, we noted that genes upregulated in ICI-2 were enriched in pathways related to bacterial infection of epithelial cells. This finding suggested that microbes in ICI-2 were involved in modulating the immune system. The microbial species richness and diversity in ICI-2 were significantly higher compared to the other two immune subtypes. We identified 17 species that were relatively enriched in ICI-1, 29 in ICI-2, and 8 in ICI-3. Notably, five microbial species were found to be specifically expressed in ICI-2. Among these five specific microbial species,
Porphyromonadaceae bacterium H1 warrants particular attention. Previous studies have demonstrated an increased abundance of some members of the
Porphyromonadaceae family in colorectal cancer, which may influence the TME and relate to the tumor’s immune response [
29]. The role of these bacteria in the TME could involve altering the distribution of immune cells within the tumor and promoting tumor cell growth [
30]. However, research in this area is ongoing, and the specific mechanisms by which
Porphyromonadaceae bacteria affect tumor development and treatment are not yet fully understood.
The advantages of machine learning in selecting microbial features are primarily manifested in its ability to process and analyze large-scale, complex microbiome datasets. This approach identifies microbial biomarkers associated with specific health conditions or diseases. Machine learning efficiently handles the high dimensionality and complexity of microbiome data, uncovering intricate interactions between microbes and their hosts [
31]. Furthermore, machine learning models can predict disease risk or treatment responses by learning patterns in the data, thereby supporting personalized medicine. For instance, in cancer research, machine learning is utilized to analyze the relationship between gut microbiomes and cancer progression [
32]. It is also used to identify microbial communities associated with oral diseases [
33]. These applications demonstrate the immense potential of machine learning in microbiome research. In this study, we initially employed a random forest algorithm to select immune-related microbial features with predictive value for differentiating long-term and short-term survival. Subsequently, using these six microbial features, we constructed a predictive model utilizing a neural network algorithm. The robust performance of our IRM model, particularly in comparison to the traditional TNM staging model, highlights the value of integrating microbial data into cancer prognosis. The significant predictive accuracy in both short-term and long-term survival predictions, validated across training and testing datasets, reinforces the model’s relevance in clinical settings. Moreover, the decision curve analysis confirms the practical advantage of the IRM model, suggesting its efficacy in guiding treatment decisions more effectively than conventional methods. These results pave the way for a more nuanced understanding of HNSCC prognosis and potentially open new avenues for personalized treatment strategies that consider the unique microbial composition of each patient’s tumor. Ultimately, the integration of multi-omics data, including microbial features, into clinical practice could significantly enhance patient management and treatment outcomes in head and neck cancer.
This study has several limitations. For instance, as indicated by the transcriptome data analysis, a single taxon might show associations with different immune-cell subtypes in conflicting ways. Secondly, we were unable to delineate the functional characteristics of each immune-related microbe and their interaction with immune cells. To validate the role of microbes more conclusively in affecting immune-cell infiltration, further in vitro and in vivo experiments are necessary.
In conclusion, this study offers a comprehensive exploration of the ICI landscape in HNSCC. We provide a detailed scenario of immune regulation in HNSCC and, for the first time, report a correlation between differing ICI patterns, the intratumor microbiome, and patient prognosis. This research aids in identifying prime candidates for optimizing treatment strategies in HNSCC.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.