Zuma 7 has OS benefit... Not so fast.
Poor subsequent therapy and a select population appears to plague this trial.
Kite has made an announcement. Zuma-7, a randomized control trial of axicel versus auto transplant, has achieved an OS benefit.
Let's be clear on who's in this study. It's patients with DLBCL who are primary refractory or those who relapsed within 12 months. Relapse within 12 months is not the same as relapse at 24 months. Moreover, we all know there's a big difference between relapse at 2 months and just at the 12-month cutoff. I have not found data on how many relapsed at 0-2, 2-4, 4-6, etc. months. Does anyone have it?
In other words, these are the patients who, a-priori, you would think would be least chemo sensitive. Thus Kite’s headline is inaccurate (FYI investors). It should say for primary refractory and relapsed within a year.
Before this trial was published, the standard of care for these patients was debated. Some doctors took these patients straight to CAR-T. Others would attempt auto, but if that didn't work, they would go to CAR T.
How do you decide which strategy is best? Ideally Zuma 7 would tell us. You randomize people to car t or auto, with car-t reserved for failure in the control group. You follow overall survival, and global health related quality of life over the duration of the study. (Key word is over the duration, as we explained here.)
That is not what happened.
The press release makes it clear that if you were assigned to the auto transplant arm, 40.5% of those patients ultimately relapsed or required more therapy. (My count of the Consort is actually a bit higher, but we need to see the data) These people should get CAR-T as the next therapy.
Yet, only 35% of those who COMPLETED STEM CELL TRASNPLANT got CAR-T (per press release).
Wait— see those key words— who received on protocol transplant— how many was that?
It appears to be 62 of the initial 179. So am I to believe that only 22 people on the control arm ultimately got CAR-T, when the right answer should be way way more?
CAR-T on the control arm at progression is not CONFOUNDING and does not DILUTE the OS benefit. CAR-T on the control arm at progression is AN ETHICAL NECESSITY, and MAKES THE TRIAL CONSISTENT WITH CURRENT PRACTICE and not having it is BAD.
The Press Release states: “At the time of the primary EFS analysis, more than half of patients in the SOC arm subsequently received Yescarta off study.”
There are 3 points here
All patients who come off for PD and are fit should get CAR-T
They should not get it eventually they should get it as immediate subsequent therapy
The company should have built crossover into the trial to ensure it happens. (They have access to the product, it turns out)
A study author says this about the press release
Unfortunately, the comment is inaccurate.
The burden to end the “try chemo first argument” would be to show that CAR-T has better OS or HRQoL vs auto followed by free and fair access to CAR-T as immediate subsequent therapy.
There are not enough details presented yet to know if that happened, and the press release suggests it likely did not occur.
Let us not forget Kite will benefit from a trial of CAR-T vs auto with no great subsequent therapy. But that is not the clinical question. CAR-T was already approved in latter lines, and clinical practice already embraced it, particularly in the subset of relapsed people with primary refractory disease or relapse within 12 months.
Trials must be conducted with control arms that reflect current practice patterns. This is well recognized by European health technology agencies, the last bulwark against flawed and limited studies.
Thought provoking as always!
The English HTA body NICE has actually published its draft recommendations on axicel for r/r DBCLC in 2L. It largely accepted the comparator from ZUMA-7 because axicel had not, at the time of assessment, been recommended for “routine commissioning” in the NHS (it has now, as it happens). They therefore allowed the company to adjust for crossover in secondary analyses.
The company also submitted additional data to NICE and state that 56% of patients in the SoC arm went on to receive CAR-T - so much higher than the 22 you reasonably estimated. The companies submission (plus academic critique) is available.
See NICE decision summary here: https://www.nice.org.uk/guidance/gid-ta10580/documents/129
And company submission to NICE here: https://www.nice.org.uk/guidance/gid-ta10580/documents/committee-papers