Question 19 of 20 Advanced Interpretation
What evidence would persuade you that an immunotherapy response is durable rather than transient?
Short answer
Look for responses confirmed on repeat imaging and maintained over long follow-up, judged with criteria built for immunotherapy such as iRECIST, which demand confirmation of progression. Support it with time-to-event data such as duration of response, and with biology showing persistent immune activity.
Why
Immunotherapy produces unusual patterns. Tumors can grow or show new lesions early, from immune infiltration or delayed effect, and then shrink. That is pseudoprogression. iRECIST therefore labels first progression as unconfirmed, iUPD.
It is confirmed, iCPD, only if the next scan 4 to 8 weeks later shows further growth, for example at least 5 mm more in target lesions. If the tumor shrinks instead, the bar resets. A single scan is not enough either way.
Durability is shown by duration of response with enough follow-up and patients at risk, not by response rate. A plateau at the tail of the progression-free survival curve, with many patients still responding years later, is the characteristic signal. Biology adds support, such as persistence of tumor-specific T cells or clearance of circulating tumor DNA.
Caution is needed. Small single-arm studies, short follow-up and censoring can make responses look durable. Responses after treatment stops are particularly informative, because the effect then cannot come from ongoing drug exposure.
A strong answer covers
- Confirmed responses on repeat imaging
- iRECIST confirmation of progression 4 to 8 weeks later
- Duration of response and survival tails with adequate follow-up
- Supporting biology such as persistent T cells or ctDNA clearance
- Responses maintained after treatment stops
Common mistakes
- Calling early tumor growth on one scan definite progression
- Judging durability from objective response rate alone
- Ignoring censoring and short follow-up when reading a duration-of-response curve
The interviewer might ask next
How would you handle a patient who meets iUPD but feels clinically well?