ESTRO 2025 - Abstract Book

S1125

Clinical – Head & neck

ESTRO 2025

also present in the human oral microbiota, have shown potential to mediate immune response and to modulate cancer treatments [2]. Recently, a 54-gene combined hypoxia/immune signature for HNSCC was developed and shown to be prognostic [3]. However, a consistent fraction of patients analysed in the study showed a mixed expression patterns and remained unclassified. To resolve the risk stratification of the mixed cluster of patients, we analysed and compared the bacterial composition of tumours with hypoxia-high/immune-low and with hypoxia-low/immune-high features. Material/Methods: The 54-gene hypoxia/immune transcriptomic signature was used to stratify HNSCC cancer patients from the TCGA data set based on their normalised mRNA profiles (hierarchical clustering, Euclidean distance, Ward method) [4]. Microbiome data (genus level relative abundance tables was obtained) for a subset of 155 HNSCC patients from the TCMA database [5]. Taxa relative abundances were centre-log-ratio transformed after replacement of zeros via Geometric Bayesian multiplicative method and limma was used to select genera differentially abundant between hypoxia-high/immune-low and hypoxia-low/immune-high tumours. A tree model was then trained to classify patients in the two hypoxia/immune clusters only on the basis of the relative abundances of selected genera. TCGA metadata and survival data were used to assess the prognostic power of the tree model in stratifying the mixed group of patients in survival analysis. Results: N=519 HNSCC cancer patients were hierarchically clustered into hypoxia-high/immune-low (n=146), hypoxia low/immune-high (n=111), and mixed (n=262) groups using the signature. Analysis of intra-tumour microbiome data available for a subset of n=155 non-metastatic patients revealed six genera differentially abundant between hypoxia-high/immune-low (n=41) and the hypoxia-low/immune-high (n=41) groups (|logFC|>1 and p<0.05), including Rothia , Peptostreptococcus , Actinomyces , Tannerella , Lactobacillus and Bifidobacterium (Figure 1B).