KEY TAKEAWAYS
- The study aimed to develop a SVM classifier using DNA methylation profiles to identify primary tissues in hnCUP.
- DNA methylation in hnCUP resembles primary tumor tissue cancers metastasizing to cervical lymph nodes, aiding accurate prediction.
The lack of tissue origin identification in head and neck cancer of unknown primary (hnCUP) necessitates invasive diagnostics and non-specific treatment. Squamous cell carcinoma, the predominant subtype, arises from diverse sites like the oral cavity, oropharynx, larynx, head and neck skin, lungs, and esophagus.
DNA methylation, highly tissue-specific, is effective in classifying tissue origin. Therefore, Leonhard Stark and the team aimed to construct a support vector machine (SVM) classifier using publicly accessible DNA methylation profiles from squamous cell carcinomas (n = 1103) commonly metastasizing to cervical lymph nodes.
This classifier aimed to determine the primary tissue origin of the cohort’s squamous cell hnCUP patient samples (n = 28). Methylation analysis utilized the Infinium MethylationEPIC v1.0 BeadChip by Illumina.
The SVM algorithm demonstrated the highest overall accuracy among tested classifiers, achieving 87%. Squamous cell hnCUP samples exhibited DNA methylation patterns resembling those of squamous cell carcinomas frequently metastasizing to cervical lymph nodes.
The most commonly predicted cancer sites were the oral cavity in 11 cases (39%), followed by the oropharynx and larynx (both 7, 25%), skin (2, 7%), and esophagus (1, 4%). These frequencies align with expected lymph node metastasis distributions from epidemiological studies.
The study showed that on the DNA methylation level, hnCUP resembles primary tumor tissues commonly metastasizing to cervical lymph nodes. The SVM-based classifier accurately predicts the tissues of origin for these cancers, potentially reducing the invasiveness of hnCUP diagnostics and enabling more precise therapy upon clinical validation.
Open access funding organized by Projekt DEAL.
Source: https://pubmed.ncbi.nlm.nih.gov/38528631/
Stark, L., Kasajima, A., Stögbauer, F. et al. “Head and neck cancer of unknown primary: unveiling primary tumor sites through machine learning on DNA methylation profiles.“ Clin Epigenet 16, 47 (2024). https://doi.org/10.1186/s13148-024-01657-3.
