Molecular Profiling in Carcinoma of Unknown Primary Site

Molecular profiling has transformed the diagnostic and therapeutic landscape for carcinoma of unknown primary site (CUP), a heterogeneous group of metastatic malignancies in which the site of origin remains undetermined despite extensive clinical work-up. Traditional approaches such as imaging and immunohistochemistry often fail to resolve the tissue of origin, leading to empirical treatment with limited efficacy. Advances in high-throughput sequencing and epigenomic analysis now enable interrogation of DNA methylation patterns, gene-expression signatures and genomic alterations in both tumour tissue and circulating cell-free DNA (cfDNA). Integration of these data by machine learning classifiers can predict tissue of origin with high accuracy, refine prognosis and identify actionable targets for personalised therapy. Liquid biopsy approaches allow minimally invasive sampling, overcoming barriers of tissue availability and permitting serial monitoring. Collectively, these developments offer the potential to convert CUP from an orphan diagnosis into a precision-medicine paradigm, improving patient stratification and guiding site-specific or molecularly targeted treatments.

Research from Nature Portfolio

Recent studies have demonstrated the power of cfDNA methylation profiling to resolve tissue origin in CUP cohorts. A machine learning classifier trained on methylation patterns across multiple cancer types achieved near-complete accuracy in predicting the primary site, even from low-volume plasma samples, and showed robust performance when combined with mutational data. This approach not only predicted tissue of origin in over 95% of cases but also led to diagnostic reclassification and informed treatment decisions in a substantial proportion of patients. These findings underscore the clinical utility of liquid-biopsy methylation assays in overcoming challenges of tissue heterogeneity and sampling bias, and they pave the way for broader adoption of cfDNA-based molecular diagnostics in CUP management.

Molecular Profiling in Carcinoma of Unknown Primary Site publication trend

The graph below shows the total number of articles in molecular profiling in carcinoma of unknown primary site across all publications each year (not limited to Nature Index journals).

Technical terms

Carcinoma of Unknown Primary (CUP): A metastatic cancer in which the anatomical origin cannot be identified after standard clinical and pathological evaluation.

Circulating Cell-Free DNA (cfDNA): Fragments of tumour-derived DNA released into the bloodstream, accessible via liquid biopsy for non-invasive molecular analysis.

DNA Methylation: An epigenetic modification involving the addition of methyl groups to cytosine residues, influencing gene expression and serving as a tissue-specific signature.

Transcriptome: The complete set of RNA transcripts produced by a cell or tissue, reflecting gene-expression patterns used for molecular classification.

Machine Learning Classifier: A computational algorithm that learns from training data to assign diagnostic labels, such as tissue of origin, to new biological samples.

References

1. A cfDNA methylation-based tissue-of-origin classifier for cancers of unknown primary. Nature Communications (2024).
2. Immunohistochemistry for Diagnosis of Metastatic Carcinomas of Unknown Primary Site. Cancers (2018).
3. Evaluating DNA Methylation, Gene Expression, Somatic Mutation, and Their Combinations in Inferring Tumor Tissue-of-Origin. Frontiers in Cell and Developmental Biology (2021).
4. Application of a Neural Network Whole Transcriptome–Based Pan-Cancer Method for Diagnosis of Primary and Metastatic Cancers. JAMA Network Open (2019).
5. Molecular characterisation and liquid biomarkers in Carcinoma of Unknown Primary (CUP): taking the ‘U’ out of ‘CUP’. British Journal of Cancer (2018).