There are, depending on the moment in which you take the measurement, 100% measurable lesions after SRS. That is to say, get an MRI the day after SRS, 100% of those SRS'd lesions are still hangin' around. Now 90+% of them are "dead men walking lesions" and doomed to regress... but a fraction aren't. An unpredictable fraction. But send all your SRS patients over the med onc the day after SRS, and if he/she gets an MRI on the patient... "Sir you still have a brain met and I want to give you keytruda."
Interesting interpretation. Let me twist it a bit further.
Mr. Miller has a brain met from NSCLC. Mr. Miller get's SRS.
Mr. Miller gets immunotherapy, since his brain met is still visible in the MRI 4 weeks post SRS.
Mr. Smith has a brain met from NSCLC. Mr. Smith gets surgery, followed by SRS to the cavity.
Mr. Smith does not get immunotherapy, since his brain met is not visible in the MRI 4 weeeks post surgery/SRS.
Does that make sense?
I'm being completely silly in my analogy but you get the point. There will be some SRS patients well after the SRS who have measurable brain mets and the med oncs will want to give them immuno specifically for the brain mets.
Brain mets on early MRI post SRS behave differently in a patient to patient basis without clear clinical implications. If you perform an MRI 4 weeks post SRS you may find the same brain met you treated before, you may find a bigger brain met, you may find a brain with less contrast enhancement, you may find no brain met. Yet, all these patients usually have the same prognosis, since MRI 4 weeks post SRS usually says little about the vitality of the brain met. In 99% of cases you will rescan in a patient without symptoms and not draw ANY therapeutic consequence irrelevant of the scan's result. The reason we do the stupid scan is to make sure another brain met has not popped up. But the scan itself is more or less useless at 4 weeks for the brain met that was treated.
We know keytruda works for some NSCLC brain met patients (and has a response rate of x%), and we know SRS works for some brain met patients (and has a response rate of y%). What I have yet to see happen in oncology (has it ever happened?) is when two separate, disparate treatments have been tried in a disease, given response rates "x" and "y" with y being greater, that x+y<y. At worst, x+y=y. But usually, almost always, x+y>y; it's been a rather predictable thing. (You can make this x+y>y argument about WBRT+SRS; but then we get into wholly valid toxicity concerns, risk/benefit ratios etc.)
I never said that Keytruda will not provide a higher response rate. It will likely do so. Yet, is radiologic response rate a valid endpoint? Is it a meaningful endpoint for the patient? Why haven't we been giving the patients chemo in the past for exact the same scenario? A cisplatinum-doublet will also provide a X response rate in the brain.
Unless you can PROVE that Keytruda will ENHANCE overall survival in patients with ONLY brain mets who are getting Keytruda post SRS of all brain mets, over giving Keytruda at progression, there is no clear indication for immunotherapy in this setting.