Precision Oncology: Why the Test Comes Before the Drug

From organ to molecule: the shift

Classical chemotherapy is a carpet bomb. It targets any rapidly dividing cell, which is why it wounds hair, gut and marrow alongside the tumour. Targeted therapy is a guided missile: it aims at a specific molecular defect that the cancer depends on — a mutated protein, an amplified gene, a receptor the tumour cannot live without. But a guided missile is useless without coordinates. The coordinates come from a biomarker (Biomarker): a measurable molecular feature of the tumour. When a test for that biomarker is legally and clinically required before a specific drug is given, it is called a companion diagnostic (Companion diagnostic). No marker, no target, no reason to fire.

Amplified receptors: HER2 and trastuzumab

Some tumours make far too many copies of a growth-signal receptor. In roughly 15–20% of breast cancers the HER2 gene is amplified, flooding the cell surface with HER2 receptors that scream "divide" without stopping. That excess is both the tumour's engine and its Achilles' heel. Trastuzumab, a monoclonal antibody against HER2, only helps when the receptor is overexpressed — so pathology must confirm HER2 positivity (by immunohistochemistry, and FISH for equivocal cases) before the drug is worth giving. The same HER2 test now guides therapy in HER2-positive gastric cancer too: the marker travels between organs.

Driver mutations: the lung and melanoma kinases

In many cancers, a single mutated kinase is the "driver" — the one broken switch that keeps the growth pathway jammed on. Block that switch and the tumour can collapse. In non-small-cell lung cancer, testing routinely looks for an EGFR mutation, an ALK rearrangement, and a ROS1 rearrangement; each unlocks its matching kinase inhibitor — osimertinib for EGFR, and ALK/ROS1 inhibitors such as crizotinib or alectinib for the fusions. A BRAF V600E mutation is the classic driver in melanoma (and some lung cancers), treated with a BRAF inhibitor plus a MEK inhibitor. The pattern never changes: find the driver, match the inhibitor.

A marker that says "don't": RAS in colorectal cancer

Some biomarkers unlock a drug; others lock it away. Anti-EGFR antibodies such as cetuximab and panitumumab block the EGFR receptor at the top of a signalling chain in colorectal cancer. But if the gene DOWNSTREAM of that receptor — KRAS or NRAS, collectively RAS — is itself mutated, the pathway is switched on from below the blockade. Blocking the receptor changes nothing, because the signal now starts past the door you just locked. So anti-EGFR antibodies work only when RAS is wild-type (not mutated). Testing RAS status is mandatory before prescribing; a RAS mutation is a firm reason NOT to give these drugs.

Synthetic lethality: BRCA and PARP inhibitors

The elegant idea here is called synthetic lethality: two repair systems that a cell can survive losing one of — but not both. A cell with a BRCA1 or BRCA2 mutation has already lost one major DNA-repair pathway (homologous recombination). PARP inhibitors block a second repair route the cell was leaning on. Knock out both and the cancer cell's DNA damage becomes unfixable — it dies, while normal cells (with intact BRCA) shrug the drug off. This is why olaparib and other PARP inhibitors are matched to BRCA-mutant ovarian, breast and prostate cancers. The mutation that helped cause the cancer becomes the exact reason a drug can kill it.

When the marker beats the organ: immunotherapy

Immune checkpoint inhibitors release the brakes on the patient's own T-cells so they can attack the tumour. But they help most when the tumour is "visible" to the immune system, and three biomarkers predict that. PD-L1 expression tells you the tumour is using the PD-1/PD-L1 brake — the very brake these drugs cut. A high tumour mutational burden (TMB) means many mutations, so many abnormal proteins for T-cells to recognize. And microsatellite instability (MSI-high) — the fingerprint of a broken mismatch-repair system (dMMR) — produces exactly that flood of mutations. Pembrolizumab is approved tissue-agnostically for MSI-high / dMMR tumours: here the marker, not the organ, decides the drug.

Two words students swap: predictive vs prognostic

This distinction is a favourite exam trap. A predictive biomarker predicts whether a specific TREATMENT will work — it guides drug choice (HER2, EGFR, RAS, BRCA, PD-L1 are all predictive). A prognostic biomarker predicts the likely OUTCOME of the disease regardless of treatment — it tells you how aggressive the cancer is, not which drug to reach for. Some markers are both, but the question you must always ask is: does this marker change WHAT I prescribe (predictive), or only how worried I am about the course (prognostic)? Confusing the two leads to giving — or withholding — the wrong drug.

The tumour no longer needs to be cut out to be read. Traditionally a biomarker meant a biopsy — a piece of tissue under a microscope. But dying tumour cells shed fragments of DNA into the bloodstream, and a liquid biopsy (Liquid biopsy) sequences this circulating tumour DNA (ctDNA) from a simple blood draw. It can find a targetable mutation when tissue is scarce, track whether a treatment is working, and catch a resistance mutation emerging before a scan would. It is an emerging complement to tissue testing, not yet a full replacement — but it hints at a future where the whole molecular profile is read from a tube of blood.

Markers that predict harm: pharmacogenomics

Not every biomarker chooses a drug — some warn you about TOXICITY before you dose. These are pharmacogenomic markers, usually inherited variants in the enzymes that clear a drug. Test them and you can avoid a catastrophe. A DPD deficiency (DPYD variant) cripples the enzyme that breaks down fluoropyrimidines like 5-FU and capecitabine, risking lethal toxicity. A UGT1A1 variant slows clearance of irinotecan, raising the risk of severe diarrhoea and neutropenia. And a TPMT (or NUDT15) deficiency means thiopurines like azathioprine and 6-mercaptopurine accumulate to dangerous levels. Same logic as the whole chapter — test first — but here the test protects the patient from the drug rather than pointing to it.