From Molecular Report to Clinical Decision

A molecular report arrives. Somewhere downstream, a treatment decision is made. Between those two points sits a chain of reasoning that is frequently compressed into a single move — "the tumour has X, therefore consider Y".

Decompressing it is useful, because the chain is only as strong as its weakest link, and the weak link is not always where one expects.

Step one: is the finding technically sound?

Allele fraction, read depth, tumour purity, specimen quality. A call at 3% allele fraction in a specimen with 20% tumour content is a different claim from one at 40% in a specimen with 70% content.

Also: was the alteration type detectable by this assay at all? A report silent on fusions from a sequence-variant assay says nothing about fusions.

Step two: is the alteration functionally consequential?

Not every alteration in a gene changes what the gene does. Position matters, alteration type matters, and the distinction between an established activating alteration and an uncharacterised missense change in the same gene is substantial.

This is where variants of uncertain significance sit. The classification is honest: the functional consequence is genuinely unknown.

Step three: is it consequential in this disease?

The same alteration can carry different weight in different tumour types. Evidence tiering frameworks encode this explicitly — an alteration with strong evidence in one disease may sit several tiers lower in another.

Tissue-agnostic approvals exist and are real, but they are the exception rather than the model. Assuming transferability by default is where a good deal of over-interpretation originates.

Step four: is it consequential in this clinical situation?

Prior treatment matters. An alteration present at diagnosis and an alteration emerging after three lines of therapy mean different things, and the second may represent selection rather than a fresh opportunity.

Disease burden, performance status, organ function and what the patient is prepared to undergo all bear on whether a molecularly-plausible option is an actually available one.

Step five: what is the strength of the therapeutic evidence?

From randomised trial data through single-arm studies to case reports and preclinical work. Each supports a different strength of recommendation, and a good interpretation states which applies rather than presenting all findings in the same register.

Step six: what are the alternatives?

A molecularly-guided option competes with standard options, with trial participation, and sometimes with the reasonable choice not to treat. The molecular finding does not automatically win that comparison, and framing it as the only relevant input is a distortion.

Why decomposing helps

Two reasons. It locates uncertainty precisely — a chain can be strong technically and weak evidentially, and knowing which changes how the finding is weighted. And it makes the reasoning transmissible: a colleague, a tumour board or the patient can see where the inference is load-bearing.

Compressed reasoning is faster and works well in familiar situations. In complex cases — which are the ones that reach molecular review — the compression is usually where the error hides.