From a molecular finding to a treatment question
Several tests can support the same biological story without proving that a drug will help a patient. A result becomes useful when it answers a defined clinical question in the right treatment setting.
Before you start: Learn what the molecular layers measure. This final lesson asks how findings travel into a decision.
Follow the claim through five gates
| Gate | Question | What still needs evidence |
|---|---|---|
| A real finding | Is the variant or measurement reproducible and from the tumor? | Sample quality, tumor purity, matched normal, and assay validity |
| A biological effect | Does it change RNA, protein, or function? | A predicted effect is not a measured one |
| A dependence | Does the tumor need this pathway to survive? | Activity alone does not establish dependence |
| A usable treatment | Can a drug reach and inhibit that target safely? | Exposure, resistance, normal-tissue toxicity, and delivery |
| Clinical benefit | Does treatment help people in this setting? | Population, comparator, outcomes, indication, and eligibility |
Independent agreement can strengthen a particular gate. It does not let a result skip later gates. Some validated biomarkers require one clinical assay; testing every molecular layer is not a prerequisite for effective standard care.
A worked example: a DNA-repair finding
Imagine a tumor with a BRCA1 alteration, reduced repair activity, and a strong response to a poly(ADP-ribose) polymerase (PARP) inhibitor in a model. These are made-up teaching observations.
Together they support a PARP-sensitivity hypothesis. They do not establish the U.S. adjuvant olaparib indication. That indication requires qualifying germline BRCA1/2 mutations and high-risk, HER2-negative early breast cancer after chemotherapy. A somatic alteration or homologous-recombination-deficiency (HRD (homologous recombination deficiency)) score does not substitute. FDA approval.
A metastatic trial, an expert-reviewed off-label discussion, and approved adjuvant treatment have separate evidence and eligibility. See The repair evidence layers.
Four tempting shortcuts
- “The protein is abundant, so its drug is indicated.” Expression supports target biology. An antibody–drug conjugate still needs the relevant clinical indication; research RNA or proteomics cannot replace required clinical pathology.
- “One checkpoint antibody scored higher, so switch drugs.” A single unvalidated ex vivo result cannot establish clinical superiority. Technical variation, viability, exposure, and culture conditions must be assessed before proposing a biological explanation.
- “ctDNA is detectable, so ablate a small lesion.” Circulating tumor DNA (ctDNA) does not locate a lesion. Confirmation, imaging, pathology, disease extent, and clinical evidence remain separate requirements.
- “No ctDNA was found, so recurrence risk is a fixed low percentage.” A negative sample means no signal detected under that assay's conditions. It is not a personal forecast or evidence that indicated treatment can be omitted.
Try it
A report shows increased AKT-pathway RNA and names an AKT inhibitor. Which gate has it reached?
Answer: it may suggest activity. It has not yet demonstrated protein activation, dependence, drug sensitivity, or clinical benefit in this treatment setting.
Explain it back
“Agreement between tests strengthens ______, while a treatment choice also requires ______.”
One answer: “a specific biological claim; clinical evidence, eligibility, and a benefit–risk assessment.”
Takeaway
Ask which claim each test supports and which gate remains untested.
Applied to Diana
Current application belongs in Diana's molecular profile, the treatment plan, and the decision pages. Ordering and timing belong in the test tracker.
Sources and scope
Source check: October 8, 2026. This lesson teaches reasoning, not a treatment prescription.
- FDA: adjuvant olaparib.
- ESMO ctDNA recommendations.
- Reading cell-therapy evidence — keeping evidence rungs separate.