Clinical Evidence: How We Assess It
The hierarchy we use to weigh supporting data, the frameworks we rely on, and why confidence should never exceed the strength of the evidence behind it.
Every recommendation in precision oncology rests on evidence. The challenge is that evidence exists at very different levels of certainty.
A laboratory experiment, a case report, a phase I trial, a randomized trial, and an international guideline can all be cited together in support of the same molecular alteration, but they should not carry the same weight. The role of molecular interpretation is not only to gather evidence; it is to assess how much confidence that evidence deserves.
The role of molecular interpretation is not only to gather evidence; it is to assess how much confidence that evidence deserves. Our goal is not to justify a treatment, but to calibrate confidence.
Why evidence assessment matters
Most interpretation errors do not arise from a lack of evidence. They arise from overstating it.
Most interpretation errors do not arise from a lack of evidence; they arise from overstating it.
Common examples include:
- Presenting a preclinical observation as if it were clinical evidence.
- Treating an off-label rationale as equivalent to a standard-of-care indication.
- Assuming evidence from one tumor type automatically applies to another tumor.
- Ignoring uncertainty when evidence is conflicting.
The quality of a recommendation depends not only on what evidence exists, but on how honestly its limitations are acknowledged.
The IYGAS evidence pyramid
Not all evidence contributes equally to clinical decision-making. We assess supporting data using a structured hierarchy, and moving down the evidence pyramid, uncertainty increases.
Proven clinical benefit Level 1
- Regulatory approvals
- International guidelines
- Positive randomized trials
- Prospective validation studies
These findings form the basis of standard clinical practice.
Strong clinical evidence Level 2
- Prospective non-randomized trials
- Consistent multicenter data
- Large observational cohorts
These can influence clinical decisions but often require additional context.
Emerging clinical evidence Level 3
- Retrospective studies
- Small patient series
- Early clinical signals
These generate hypotheses but should not be read as proven benefit.
Anecdotal evidence Level 4
- Case reports
- Individual patient experiences
These observations can be informative but rarely justify treatment decisions on their own.
Biological rationale Level 5
- Preclinical studies
- Cell line models
- Animal experiments
- Mechanistic hypotheses
These help explain biology but do not demonstrate clinical benefit.
Tumor-matched evidence leaves tumor-agnostic assumptions behind
One of the most common errors in molecular oncology is assuming that evidence supporting an alteration in one cancer automatically applies to another. A biomarker may be highly actionable in one tumor type, weakly supported in a second, and clinically irrelevant in a third.
Wherever possible, we prioritize evidence generated within the same tumor type. The question is never:
“Is this alteration actionable somewhere?”
The question is:
“Is it actionable here?”
The frameworks we use
Clinical interpretation must be reproducible. That is why we ground recommendations in established evidence systems.
Variant classification AMP · ASCO · CAP
Used to classify the clinical significance of somatic variants.
Clinical actionability ESCAT
The ESMO Scale for Clinical Actionability of molecular Targets; ranks alterations by the strength of evidence supporting their clinical benefit.
International guidelines NCCN · ESMO · ASCO
NCCN, ESMO, ASCO, and tumor-specific expert consensus recommendations.
These frameworks provide the foundation. Clinical judgment is applied on top of them, not in place of them.
Biological plausibility is not clinical evidence
Some molecular findings have a compelling biological rationale but limited clinical evidence. A pathway may be clearly activated, a resistance mechanism may seem plausible, a targeted therapy may work in laboratory models. These observations can be valuable; they may justify further research, enrollment in a clinical trial, or a careful off-label discussion. But biological plausibility should never be presented as proven clinical benefit.
We draw a clear distinction between four levels:
- Confirmed evidence, supported by robust clinical data.
- Emerging evidence, supported by early clinical observations.
- Biological hypothesis, supported primarily by mechanistic rationale.
- Speculative interpretation, interesting but currently unproven.
Maintaining these distinctions is essential to honest reporting.
How we approach off-label opportunities
Off-label recommendations are not automatically inappropriate; many important advances in precision oncology began as off-label observations. But they must be evaluated with the same discipline applied to standard treatments. For every off-label option, we ask:
- Is the biological rationale convincing?
- Does clinical evidence exist?
- What level of uncertainty is present?
- Are clinical trials available?
- Is there a better-supported alternative?
An off-label idea should never be presented as if it were a guideline-supported standard.
From evidence to confidence
The purpose of evidence assessment is not to determine whether a treatment is possible. It is to determine how much we should trust it. This philosophy is reflected in the confidence indicators used throughout the IYGAS report: strong evidence earns strong confidence, weak evidence earns appropriately limited confidence.
Confidence should never exceed the strength of the supporting data.
Precision oncology is not about finding a rationale for treatment. It is about understanding how strong that rationale actually is. Evidence does not eliminate uncertainty; it helps define it. That is why the goal of evidence assessment is not certainty but transparency; because better decisions begin with an honest understanding of what is known, what is uncertain, and what remains to be proven.
This content is for educational purposes only. It does not constitute individual medical advice and does not replace the judgment of the treating physician.
Submit a case for expert molecular review
NGS Advisory reviews your existing NGS report against current guidelines, clarifies the exact variant and its actionability for your tumor type, and delivers a structured, calibrated second opinion.