Expert Financial Analysis and Reporting

Reason for This Further Discussion

A number of subscribers have pointed out to me that my discussion of the survival tail could have been more clearly explained. Consequently, I have rewritten two sections as shown below that will hopefully be clearer. I have also summarized my analysis as follows:

• In the CheckMate trials in second line non-small cell lung cancer, 13.4% of patients treated with the checkpoint inhibitor Opdivo survived for 5 years versus 2.6% for the previous standard of care, the chemotherapy drug, docetaxel. This delta of 10.8% is the survival tail and these types of results in aggressive cancers were the driving force that resulted in the enormous commercial success of Opdivo and the nearly identical product Keytruda.
• The five year survival rate for current standard of care in newly diagnosed GBM is estimated at 5%. This indicates that newly diagnosed GBM and second line non-small cell lung cancer are comparably aggressive. If 15% of newly diagnosed GBM patients treated with DCVax-L are alive at 5 years, it would have a delta of 10% that would be a survival tail of comparable to Opdivo in second line non-small cell lung. I would consider this as a home run result.
• Based on Kaplan-Meier projections, 15% of all 331 patients treated in the phase 3 trial of DCVax-L in newly diagnosed GBM have survived or are projected to survive for 5 years. Relative to the 5% that would be expected to survive for 5 years if treated with SOC, the delta is 10%. Hence, there appears to be an impressive survival tail for the trial as a whole. This also would mean that the delta for DCVax-L would be greater than 10%, but how much?
• My assumptions indicate that 45 of the 50 patients projected to survive five years received DCVax-L at some point in the trial.
• If all 45 patients came from the 232 patients that received DCVax-L plus SOC at the start of the trial, this would indicate a 5 year survival rate of 19% and a highly impressive delta of 14%.
• If these 45 patients came from the 298 patients who received DCVax-L at some point in the trial, the five year survival would be 15% for a delta of 10%.
• Before you chalk up a staggeringly impressive win for DCVax-L, let me remind you that the Kaplan Meier data could get either better or worse as they continue to mature. I would also point out that one of my most critical assumptions is that 5 year survival for SOC will be less than 5%. Most KOLs agree with this assumption, but this is primarily based on their clinical experience. There is only sparse data from clinical trials.

The rewritten sections are as follows:

So What Can We Conjecture about the Possible Survival Tail of DCVax-L in Newly Diagnosed Glioblastoma?

We know that the two lead investigators on the trial, Drs. Linda Liau and Keyoumars Ashkan believe that patients in the trial are living meaningfully longer than expected. See this link. We also have a valuable insight into the trial based on an analysis of the blinded data from the trial. There were 331 patients treated in the trial; they were randomized 2:1 to DCVax-L plus SOC versus SOC. Patients started on SOC were allowed to crossover to DCVax-L if their cancer progressed. Also, patients who progressed on DCVax-L were allowed to stay on the drug. The Company has reported that ultimately about 90% of the patients received DCVax-L either initially or after crossover. So, about 232 patients were started with DCVax-L plus SOC, about 66 were started with SOC and were then given DCVax-L if their cancer progressed and roughly 33 received just SOC. Standard of care is surgical resection followed by radiation and the chemotherapy drug temozolomide.

The trial remains blinded so that investigators and others (including the Company) don’t know how patients treated with DCVax-L plus SOC fared versus those treated with just SOC. However, blinded data released on November 19, 2018 provided some very encouraging data. It was reported that for the 100 longest survivors in the trial, the Kaplan Meier estimate of median overall survival was 58.4 months or 4.9 years (let’s call it 5 years) from time of surgery. The Kaplan Meier algorithm is used universally throughout the pharma and biotech industry. It incorporates both the actual data on patients who have already reached a certain time point (in this case 58.4 months) and also predictive data on the patients who have not yet reached that time point. The Kaplan Meier median overall survival estimate indicates that at least 50 (15%) of patients in the trial have survived or are projected to survive to at least 5 years.  We don’t know how many of these 50 patients were started on DCVax-L plus SOC, how many were started on SOC and had DCVax-L added when their cancer progressed and how many received only SOC. Also, as the Company has expressly pointed out in each of its announcements about interim clinical trial data, the Kaplan Meier data could get either better or worse as they continue to mature.

How Do We Interpret The Kaplan Meier Data?

Let me first reiterate that the Kaplan Meier data is an estimate, albeit a very important one that is broadly embraced by the medical community and regulators. Based on historical data from other GBM clinical trials and clinical experience of key opinion leaders like Linda Liau, there is a general consensus that roughly 5% of patients treated with SOC will survive 5 years. If the phase 3 trial data continue to remain approximately the same as they mature, there would be 50 patients or 15% of the 331 enrolled in the trial that survived for five years or more. Remember, this includes all patients including those who were started on SOC. The expectation is that for patients who receive only SOC, 5% would be alive at 5 years so that of the 99 patients started on SOC only 5 would be expected to survive for 5 years. Of course, 66 of these 99 patients were given DCVax-L when their cancer progressed. I think that we can reasonably infer that these 66 patients might do less well than those started on DCVax-L plus SOC. If these two premises are correct, this would mean that more than 15% of patients started on DCVax-L plus SOC have survived or are expected to survive longer than 5 years. This indicates a robust survival tail for DCVax-L in this trial that is comparable to that seen with Opdivo in second line non-small cell lung cancer.

We can take this analysis further by again assuming that the five year survival for patients treated with SOC is 5%. If so, 5 of the 99 patients who started on SOC might would be expected to survive for 5 years. Remember that about 66 of these patients received DCVax-L when their cancer progressed. Some of these might also reach 5 years survival. With this assumption, we can assume that 45 of the 50 patients who have survived or who are projected to survive for 5 years received DCVax-L at some time in their treatment. This would mean that 45 patients came from the 232 patients who received DCVax-L at the start of the trial and the 66 patients who were started on SOC and had DCVax-L added when their cancer progressed. Of these 298 patients who received DCVax-L at some point in the trial, 15% have survived or are expected to survive for 5 years. If we attribute the 45 survivors only to the 232 who started on DCVax-L then the five year survival rate would be 19%. It is reasonable to infer that the 66 patients for whom DCVax-L was added when their cancer progressed would do less well than those started on DCVax-L.