1. What is actually tested
DNA is extracted from tumour tissue or plasma containing circulating tumour DNA; RNA may be needed for expression and fusions. The laboratory compares sequence to a reference and, in paired testing, to a separately collected non-tumour specimen from the patient. The result describes the sampled material and disease time point: metastases and later relapse can differ.
2. Short variants, copy number and rearrangements
SNVs and indels can activate a driver or disrupt a suppressor. Amplification increases copy number, deletion removes sequence, and a functional fusion can create an oncogenic transcript or active kinase. EGFR exon 19 deletion, ERBB2 amplification and an ALK fusion are therefore distinct biological events requiring appropriate assays.
3. Composite biomarkers: MSI, TMB and HRD
MSI/dMMR reflects mismatch-repair deficiency and can be assessed by PCR, NGS or MMR-protein IHC. TMB is a platform-dependent mutation count per megabase. HRD describes homologous-recombination deficiency, but a BRCA variant, an HRR panel and a genomic-scar assay are not interchangeable.
4. Why targeted medicines become possible
When tumour growth depends on an activated protein, a medicine may block that dependency. Tyrosine-kinase inhibitors are used for selected activating EGFR, ALK, ROS1, RET, NTRK, BRAF, FGFR and other alterations. Antibodies and conjugates can depend on HER2 expression or amplification, while PARP inhibitors depend on a defined BRCA/HRD context. The exact relationship varies by tumour and jurisdiction.
5. A biomarker can exclude a treatment
Not every finding opens a treatment. Activating KRAS/NRAS in metastatic colorectal cancer predicts lack of benefit from some anti-EGFR medicines; selected secondary EGFR or ALK alterations are linked to resistance. A negative predictive biomarker can be as clinically important as a positive one.
6. Immunotherapy and genomic signals
MSI-H/dMMR and, in selected indications, TMB-high can support discussion of immune-checkpoint inhibitors. Immune response is not reducible to one number: tumour type, PD-L1, prior treatment, authorised indication and assay limitations matter.
7. A tumour finding can trigger germline testing
Pathogenic BRCA1/2, PALB2, TP53, MMR and other tumour variants can sometimes be germline. A high allele fraction is only a clue; tumour-only analysis does not establish origin. Inheritance requires a separate germline specimen, consent and genetic counselling. The result can affect surveillance and relatives.
8. Where clinical evidence ends and research begins
Actionability is strongest when biomarker and treatment are supported in the same tumour and setting. Evidence from another tumour, an early-phase study, isolated responses, a preclinical model, an unknown variant or an algorithmic combination prediction belongs to a research tier. It may support trial search but must not look like standard prescribing.
9. Why a matched medicine may still fail
A tumour contains different clones, can use bypass pathways and evolves under treatment pressure. A target may exist in only some cells, while co-alterations suppress response. A molecular match can strengthen a treatment discussion but cannot guarantee benefit.