Can 214/nivo compete against pembro/axitinib in rcc? I am not convinced with early 214 data myself but reserve the right to change my opinion. The pembro/axitinib phase 3 will readout in 2019 I believe.

Dr. Atkins. Axitinib + pembrolizumab in 1st line #kidneycancer. Unprecedented 73% ORR. Only 3/52 pts with PD as best response, only 6 pts have died, 2 deaths unrelated to RCC. OS curve staying high at 17 mo med f/up #GU18

1 Like…

Take a look at cabo/nivo data in previously treated rcc patients.

Among the 13 patients with mRCC who were evaluable for response, ORR was 54 percent (7 PRs of 13 patients) and the DCR was 100 percent.

No treatment naive patients in rcc cohort.

214 is obviously active, but I wonder if bmy is betting on the safety profile of 214 will make it a winner. 214 comparable in efficacy, but early data doesn’t appear superior to other drug combos.

We’ll see when data matures to the point we can actually get long term OS and pfs data. Then we can really compare it to other drug combinations.


" This is an insurance policy for Opdivo, and a smart one, at that."

I think your right, bmy had too much opdivo/ipi eggs.

Joeflow is a chemist and I happen to agree with his comment seen here.…


I’m not suggesting in any way that 214 doesn’t work. And of course, these patients would die if you did nothing. One of the counterpoints I was making was that I was cautioning against giving credit to the late responder effect.

I understand that this is something not commonly seen with other onc drugs. Most conventional drugs are directly cytotoxic. Once therapeutic levels are reached it becomes largely mass vs. mass. The tumors are sensitive or resistant. If no response, you stop giving it to them.

Blockade+ 214 works in an entirely different way. It reverses T cells that are held in check, then 214 gives them the ability to proliferate to a critical mass required for a measurable response. All else being equal, if patients have high numbers of pre-existing tumor-reactive T cells that are released from being checked, they reach the critical mass earlier. If they start with lower numbers of T cells that are being checked, they respond later.

So in the two scenarios above, I would not expect late responders with a directly cytotoxic drug, but would with a drug that required an indirect amplification step. The amplification step is exponential, and the rate of cell division is fixed by other biological effects. Even in a highly controlled system, a very tiny difference in baseline levels of target T cells leads to a major change in the time it takes to reach the critical mass required to measure an effect. This is not good or bad. It’s a fundamental reflection of mechanism, and one that occurs in ANY adaptive immune system and to any antigen. So my point was to be cautious about using it to bolster a conviction that this finding in and of itself speaks to effectiveness. The overall response fraction itself compared to opdivo alone is impressive enough that it doesn’t need to added to a list of reasons to be encouraged. I can almost assure you that no FDA panel member who would be on the fence about approval is going to be swayed by the appearance of late responders. It is not uncommon with Ipi, TIL, and IL-2 alone, so I am neither encouraged nor discouraged by the observation. The overall response fraction is encouragement enough for me.

At the level of the individual patient, the finding is meaningful. If you don’t respond early, you still have hope that you might respond later.



Doc, I was also taught to develop a hypothesis, test it, and not to expect a certain outcome. I’m not saying the outcome of the trial was expected. I am saying the distribution pattern of the outcomes relative to time in a series of observations is different for an indirect amplification process than the distribution of responders over time in drugs that have a direct effect. It wouldn’t matter what the final percent responders where. It could be 10% or 90%, but the distribution pattern of individual outcomes among the responders would be different.

So if a company claims that every time they make an estimate of the fraction of responders that it goes up, I would still only draw conclusions about the drugs effectiveness at the end of the study. If it is 90% effective overall, I don’t attach more value to an individual response that occurred later in the study than any other data point. But yet that is precisely what one is doing if they attribute value to a late events. It’s like double dipping. If drugs A and B both give a 90% response rate and Drug A gets to 90% at 60 days, and drug B has 80% at day 60 and 90% at day 75, you wouldn’t say that you are more encouraged about drug B because its response rate rose over the last 15 days would you? You also wouldn’t discredit Drug A because it’s response rate had plateaued. You don’t deserve a double dose of credit for it being 90% effective, plus the fact that the response fraction was still increasing towards the end of the trial.

This has zero to do with expecting anything about the outcomes generated while testing my hypothesis, only what the distribution of data over time would look like, whether it supports my hypothesis or not.