positive interim data from can003 trial indeed

When CAN-003 interim data were presented back in Oct last year I was flat busy and had only a brief look at the data, the bar charts presented in the poster did not show any break thru, furthermore the sp slide down was shouting loud the data are below average.

Few weeks later I had time to look at the data in more details, worked out the charts back to numbers and run the results thru the stats models for survivals to estimate the positiveness. There were two main outcomes of these exercises:
1. The calcs (CI, chi-squire test, p-value) were saying the results are not positive, but nor negative (keeping in mind I did just an estimate having small sample of only progressed patients) In other words we can say it’s too early in the trial to make any long shot conclusions.
2. Reviewing the data in details generated a whole bunch of questions. I sent all my questions to ML and he kindly and thoroughly answered half of them :) (but covered all re. the trial design). The main theme was it’s too early to analyse the data.

Last weekend I was discussing Prima, CAN-003 interim results, replies from ML and etc. with someone and decided to reread the replies which prompted me to play with the data and graphs. Combining the answers from ML and sketching the data to show the trends made me change my opinion and agree with Prima the results are positive and shape a favourable trend INDEED.

Before I share that info two important things to note:
1. Prima calculates the PFS from all patients (progressed and censored) – this is not right from stats point of view, I used only progressed patients.
2. Prima used bar charts to present a snapshot of the trial data separating the 1st and the 2nd remission groups , I used boxplots which are part of EDA analysis on combined data. The boxplots are formed from summary stats that captures the main features of the data helping to gain overall picture.

Here are some important replies from ML before we plot the data:

“Your assumption is correct. These patients did not receive any CVac dosing and that these patients progressed very early. At the final analysis next year, we present 2 data analyses – the “intent to treat” and the “per-protocol” analyses. The ITT analysis includes all patients randomized, regardless of how fast they progressed. The PP analysis includes only patients that received 4 doses of CVac or who those stayed on study or more than 20 weeks (in the OSC group).


2. The protocol specified analysis does not make a distinction between 1st and 2nd remission patients for analysis; there was no stratification of randomization based on this factor. The presentation of these groups separately will be done post hoc because it is potentially interesting and because looking carefully at the 1st remission group may help us make adaptations to CANVAS if necessary. We regard these interim data as “positive” for several reasons: (a) the data are starting to look like what one expects with immune therapies for cancer, i.e. there is a time delay in the separation of the active versus control group before the active group demonstrates better performance, (b) the safety profile is extremely positive, especially as compared to any conventional cancer treatment, and (c) the ICS data indicates that CVac is having the expected effect on the cellular immune system which is consistent with anti-cancer activity. Taken all together, we believe this is positive and we expect that these interim data would continue or even improve as we continue to monitor patients.”

So if we take all the progressed records and clean the data as per PP and sketch the box plots for NRCVAC, CVAC and OSC we get the following image.

IMHO, these three plots clearly indicate to me the status of the median PFS (red circle) in all three groups and the trend in favour of CVAC group with OSC group legging behind. Another positive thing from the CVAC plot is the data are skewed to the right of the box that should support the median PFS to grow.

Furthermore you can look at the NRCVAC data as a pilot data where the CVAC group will follow the pilot. All together is positive indeed given the early status of the trial.


Add in to the picture

1. we have only 9 months of data

2. the immune response kicks in after several doses (i.e. 4) of CVAC

3. assuming average PFS is about 16-18 months (OSC) I expect the leg behind momentum increasing

4. I believe the results of 3+ months advantage of CVAC over OSC is a very good result.

-------------------------------

Nevertheless, it’s just my biased opinion and DYOR :)