AI
Sunday, October 28, 2018 7:55:35 PM
Using the updated chart, the survival for each patient was extracted; the classification (M+, M-, U) was noted; and the KM chart calculations were performed; and the results were plotted in excel. The chart below is a translucent overlay of the NWBO provided curves over the excel generated curves. As can be visually observed, the manually generated curves and the NWBO curves are almost identical.
Below is the excel chart without the NWBO overlay.
The 38 unclassified patients were analyzed separately, and the KM derived mOS was found to be approximately 19 months. Additionally, the 38 unclassified patients were removed from the blended data and the KM curve was recalculated for the remaining patients. The comparison of the KM curves is shown below:
The effect on the blended data was not as pronounced as one would intuitively think or derive using weighted average calculations. But what I find interesting is the how poorly those first 38 patients performed compared to the patients enrolled later in the trial. The poor performance is evidence (but not proof), in my opinion, that the DCVAX-L inclusion-exclusion criteria is not resulting in patients with superior survival characteristics being exclusively enrolled in the trial.
Linked below is all of the data used to generate the charts. It also contains all of the results of the KM calculations for anyone who wishes to independently verify the results.
https://drive.google.com/file/d/1KhDwssH_7Un_tUQYgQYLeMJiabyfbJaO/view?usp=sharing
I marked 33 patients as unclassified which means there were 5 (38-33) patients that lived past 36 months. And assuming the classification is accurate, this analysis supports Lykiri's supposition that several patients from the unclassified group did indeed made it past 36 months.
Below is the excel chart without the NWBO overlay.
The 38 unclassified patients were analyzed separately, and the KM derived mOS was found to be approximately 19 months. Additionally, the 38 unclassified patients were removed from the blended data and the KM curve was recalculated for the remaining patients. The comparison of the KM curves is shown below:
The effect on the blended data was not as pronounced as one would intuitively think or derive using weighted average calculations. But what I find interesting is the how poorly those first 38 patients performed compared to the patients enrolled later in the trial. The poor performance is evidence (but not proof), in my opinion, that the DCVAX-L inclusion-exclusion criteria is not resulting in patients with superior survival characteristics being exclusively enrolled in the trial.
Linked below is all of the data used to generate the charts. It also contains all of the results of the KM calculations for anyone who wishes to independently verify the results.
https://drive.google.com/file/d/1KhDwssH_7Un_tUQYgQYLeMJiabyfbJaO/view?usp=sharing
I marked 33 patients as unclassified which means there were 5 (38-33) patients that lived past 36 months. And assuming the classification is accurate, this analysis supports Lykiri's supposition that several patients from the unclassified group did indeed made it past 36 months.
