🤖 Generated with AI assistance · Does not replace individualized medical advice.

Machine Learning Identifies Immune Signature Predicting Disease-Free Survival in HNSCC After Radiochemotherapy

This post was generated with the assistance of AI, following the Oncology Hub editorial guidelines. It is not medical advice.

Key Facts

Why It Matters / Context

Head and neck squamous cell carcinoma (HNSCC) remains a challenging malignancy, with treatment outcomes heavily influenced by tumor biology and host immune response. Despite the growing interest in immunotherapy, peripheral blood-based biomarkers are not yet used to guide therapy in HNSCC PMID: 42675144. The DIREKHT study addresses this gap by prospectively integrating comprehensive immune monitoring into standard postoperative radio(chemo)therapy (R(C)T). The use of machine learning to distill a high-dimensional immune profile into a clinically relevant signature represents a step toward precision oncology. The finding that pre- and post-therapeutic immune parameters contribute equally to the predictive model underscores the dynamic nature of the immune response during treatment, suggesting that both baseline status and therapy-induced changes carry prognostic information PMID: 42675144.

Details

The study enrolled 70 patients with oral cavity or oropharyngeal cancer who received curative R(C)T between 2015 and 2026 PMID: 42675144. Peripheral blood samples were collected before therapy, after therapy, and during follow-up. Flow cytometry-based immunophenotyping assessed 45 immune parameters, including T cell subsets, monocytes, basophils, and HLA-DR expression PMID: 42675144. A machine learning workflow employed Repeated Elastic Net Technique (RENT) for feature selection within repeated stratified K-fold cross-validation, with nested cross-validation for tuning and assessment PMID: 42675144. This approach identified a 29-parameter immune signature from pre- and post-therapeutic profiles. Key contributors included HLA-DR+ T cells, HLA-DR+ monocytes, and basophils PMID: 42675144. The best model achieved a Matthews correlation coefficient (MCC) of 0.681 for predicting disease-free survival (DFS) PMID: 42675144. Adding clinical parameters (e.g., age, stage) did not improve performance (MCC 0.678), but yielded a comparable model integrating immune and clinical variables PMID: 42675144.

Limitations

The study is limited by its small sample size (70 patients) and single-center design, which may affect generalizability PMID: 42675144. The non-randomized, exploratory nature means the immune signature is hypothesis-generating rather than confirmatory. The 29-parameter signature, while predictive, is complex and may be overfitted to this cohort; validation in larger, independent cohorts is essential before clinical application PMID: 42675144. Additionally, the study did not include a control group receiving different therapies, so the signature’s specificity to R(C)T versus other modalities remains unknown. The authors explicitly state that validation in larger cohorts is required to confirm clinical applicability and reduce the signature PMID: 42675144.

Clinical Takeaway

Clinicians should consider integrating peripheral immune monitoring into HNSCC trials to explore prognostic signatures, but current evidence does not support routine clinical use.

Full Reference

Donaubauer AJ, Tomic O, Mogge L, Müller SK, Futsaether CM, Liland KH. Exploratory immunomonitoring during radiochemotherapy in HNSCC and machine-learning reveal immune parameters associated with disease-free survival. NPJ Precision Oncology. 2026;10:XX. PMID: 42675144, DOI: 10.1038/s41698-026-01658-w.