An assessment that only summarizes the data adds little. Its job is to show a decision-maker what the evidence supports, what it does not, and what would have to be true for the case to hold.
The evidence, as it was generated
Say how many patients were evaluable, how they were selected, how long they were followed and how results were reported. A response rate from a small, selected, single-arm cohort and the same number from a larger, unselected population are different findings. Note what is missing, such as durability data, central review or a clear account of patients who left the study early.
The clinical context
Place the results next to what patients receive today and what is about to change. This needs care: a cross-trial comparison cannot control for differences in population, line of therapy, assessment method or follow-up, and the assessment should say which of those differences matter most.
The safety and dose picture
Look beyond the headline adverse-event rate to dose reductions, interruptions and discontinuations. An agent that produces responses at a dose many patients cannot stay on is a different proposition from one whose dose is well tolerated, and the development plan may need to resolve that before anything else.
Where the asset is different, and for whom
Differentiation is specific: a subgroup, a line of therapy, a combination partner or a tolerability advantage. The assessment should name where the asset could plausibly change practice and which evidence would show it.
What remains to be done
Set out the studies still needed, the uncertainties that could change the plan and the points at which the program could be stopped or redirected. FDA’s guidance on endpoints for cancer drug approval is a useful reminder that the evidence needed for a regulatory conclusion differs from the evidence of activity.
Public reference points
Related service: Asset diligence.