Northwest Biotherapeutics (OTCQB:NWBO) is a clinical stage biotechnology company focused on the development of personalized immunotherapy for a broad range of solid tumor cancers. The company is developing their DCVax platform technology to treat cancer more effectively without the side effects of chemotherapy. Their lead product DCVax-L, is currently in an ongoing, late stage phase 3 clinical trial for newly diagnosed glioblastoma multiforme (GBM). While many eagerly await the final results, one is left to dissect the interim results published in the Journal of Translational Medicine, May 2018 (co-authored by 69 people, including many of the study’s principal investigators). It’s important to note that this trial has a crossover to treatment provision so that when a control patient is determined to have a progression event, they will cross over to treatment. When the treatment patient is determined to have a progression event, they remain on treatment, however their treatment schedule starts over. This crossover can confound the final data, as the JTM article has indicated that as of March 2017, 86.4% of all patients ultimately ended up on DCVax-L.
The goal at this time is to determine the likelihood of success for DCVax-L (in the treatment group), leading to FDA approval. The first thing to determine is which data will be the most important for approval. The company and the FDA had agreed at the beginning of the trial that the primary endpoint for measuring success would be progression free survival (PFS), and the secondary endpoint would be median overall survival (MOS). However, due to the recent updates on immunotherapy trials, it’s been determined that the most important data for the measurement of success is “long tail survival”. Since immunotherapy treatments have a greater effect on long term overall survival rather than the progression aspect of a clinical trial, this has developed into the “Gold Standard” , and the benchmark to measure success in immunotherapy trials. Current available data indicates for a patient with GBM on standard of care (SOC) treatment, the average survival is 15-17 months, and they have a 16% chance of surviving 36 months. I’m sure you’d agree that a patient living to 36 months with GBM would certainly be considered a long-term survivor.
Concerning long-term survival for the Phase 3 trial for DCVax-L, the blended interim results have shown that patients who have survived to 30 months have a projected survival Kaplan-Meier curve (KM-curve) of 46.5 months, but the patients living to 36 months have an even more impressive survival KM-curve of 88.2 months! The purpose of a KM-curve is to estimate one’s chances of survival if one has lived to a certain point based on a trial’s data. With the DCVax-L interim trial data, a patient who has survived to 36 months would be expected to live to at least 88.2 months. What’s even more interesting is that the data showed that this projection had almost doubled compared to a patient who had lived 30 months, as their KM projected survival was just 46.5 months. This shows that patients living to 36 months have a significantly greater chance at long-term survival. It is my belief that 36 months will be the statistical significant benchmark in this trial for measuring success. It is important to note that all of the patients in the trial as of August 2018 have been at least 36 months past surgery, and as of November 2018, they are at least 36 months past randomization.
Before getting into the success rates for DCVax-L, let’s first look at the historical survival rates for standard of care (SOC- radiation and Temodar) and the last therapy approved for GBM with the highest survival rate at 36 months (Novocure’s Optune):
- SOC- Temodar (36 month survival rate- 16%)
- Optune- (36 month survival rate – 26%)
The DCVax-L published interim paper gives us data on the whole ITT (Intent to treat, meaning all 331 patients) main arm of the trial population. Much of the data provided in the paper pertains to the 1st 182 patients in the trial. I am going to refer to this as the "1st part" of the trial (meaning the first 182 patients), and the "2nd part" of the trial (meaning the following 140 patients). Now that we have established this, lets start to dive into the numbers.
The first 182 patients were at least 36 months out from surgery at the time that data was taken (March, 2017 was the date of the interim data). We were given that the blended 36 month OS was 25.4%, which meant that of the 1st 182 patients, 25.4% lived to at least 36 months. To be clear, this includes patients from both treatment and control groups. Out of the first 182 patients, 44 had lived at least 36 months. So how do we determine how many of those began as treatment patients in the trial and received DCVax-L from the time of enrollment?
The way I approach it is by using historical survival data for 36 months for the SOC GBM patients. The most current survival data comes from the Optune study, which evidenced a 16% survival rate at 36 months.
- We know the first 182 patients in the trial were composed of a 2:1 (treatment to control) randomization, meaning that for every 3 patients randomized, 2 received treatment and 1 received control
- 182 x .33 = 60 control
- 60 x 16% = 9.6 (patients expected to live to 36 months)
- We know the first part of the trial had 44 patients make it to 36 months
- 44-9.6 = 34.4 (treatment)
- Based on that, we can guesstimate the number of treatment patients at 36 months (OS)
- 182- 60= 122 (treatment)
- 34.4/122=28.2% of treatment arm patients ≥ 36 month OS (for DCVax-L arm from the first 182 patients (the “1st part” of the trial)
Ok that sounds pretty good, what about the 2nd part of the trial? Well, it just so happens we were given a couple gifts by the interim blended blinded paper:
- The blended 36 month OS = 25.4%
- This graph shows events (represented by step-downs) and survivors (represented by up-ticks)
Source: Liau, et al. 2018 (Survival curves have been superimposed, the numbers and lines are modifications in order to make it read more clearly for purpose of calculations)
Every up-tick represents a patient currently alive at the time of analysis (3/17). This was an incredible gift! Why? Because with some statistics that were disclosed in the interim paper, one can now figure out the “attrition rate” at each month, or the likelihood of a patient surviving at a particular timeline down the graph. Example:
- Methylated patient OS% at 12 months = 94.5% (given in the paper)
- Methylated patient OS% at 24 months = 67.7% (given in the paper)
- Subtract the difference to get 27.8%
- 27.8% / 12 months = 2.32 (monthly rate)
- To determine the 6 month attrition rate, multiply 2.32 x 6 = 13.9%
- 13.9% of the 22 patients who have lived to 18 months will die before 24 months
- 86% will live to 24 months, 86% x 22= 18.9 or 19
- 12 patients were already at 24 months, add the 19 previous = 31 patients between 24 and 30 months
- Calculate the attrition rate from 24 to 30 months for these 31 patients
- Methylated OS% at 24 months = 67.7% (from journal)
- Methylated OS% at 36 months = 46.6% (from journal)
- The difference is 21.1%
- 21.1%/12 months= 1.76%
- To determine the 6 month rate, 1.76 x 6 months= 10.56% of 31 patients will die by 30 months, and 89.44% will live to 30 months
- 31 x 89.44= 27.7 or 28 will live to 30 months
- 7 patients are already at 30 months, add the remaining 28 who had survived 30 months = 35
- The same rate applies, 1.76 x 6 months = 10.56%
- So 89.44% x 35 = 31.3 or 31
- This gives us an approximate estimate of the Methylated patients from the 2nd part of the trial that should live to at least 36 months in the trial based on the journals odds of living for the 1st part of the trial, or the first 182 patients
- 44 already had lived to 36 months (given by the paper)
- 31 methylated should live to 36 months
- =75 patients expected to live to 36 months
Using the same math and statistics that were given in the interim paper, we can apply the same formula used above to determine how many of the Unmethylated patients should live to 36 months:
- 18 patients alive from 18 to 24 months
- 86.4% live to 12 months, 32.1% live to 24 months. 86.4-32.1= 54.3
- 54.3/12= 4.5% will die per month
- 4.5 x 6 = 27.18% die and 72.82% will live to 24 months
- 18 x 72.82% = 13 unmethylated patients living to 24 months
- 10 Unmethylated patients alive at 24 months already, add in the 13 who live to 24 months = 23 patients alive at 24 months
- Chance of survival at 24 months = 32.1%
- Chance of survival at 36 months = 11%
- 32.1-11= 21.1% will die and 78.9% will live to 36 months
- 23 x 78.9%= 18 unmethylated will live to 36 months
So now we have:
- 44 already had lived to 36 months
- 31 Methylated "expected" to live to 36 months
- 18 Unmethylated "expected" to live to 36 months
- 44 + 31 + 18= 93
- 93 expected to live to 36 months
- 93/331= 28.1% of entire ITT "expected" to live to 36 months
These results are very impressive, and I would be willing to argue that due to the likelihood of an improvement in attrition rates from the 149 patients that represent the 2nd part of the trial, (331-182 = 149) these results will be even better. Why do I estimate that the attrition rates will improve for the 149 patients on the 2nd part or the back end of the trial?
- It seems that the last 31 patients all received DCVax-L instead of control (the trial was halted at 300 patients with a randomization of 2:1 for treatment to control, but the final percent of patients who received treatment was 70% instead of 66.7%) The journal notes that there 232 treatment patients and only 99 control patients, when with an exact 2:1 randomization on 331 patients would have yielded 110 control patients. When the trial was resized to be 348 patients, it was intended that there would be 232 treatment and 116 control patients. After the company indicated there were only 331 patients enrolled in the main arm of the trial, most thought the breakdown would be something like 221 treatment and 110 control. Instead, there were just as many 232 treatment patients as they had originally intended, yet there were 17 fewer control patients. We know there were 300 patients enrolled by August 2015, when a screening halt was implemented - and that 31 patients were enrolled after this point. It seems logical then, that the first 300 were randomized 2:1, with 201 to treatment, 99 to control. The additional 31 patients were then all randomized to treatment. The reasons for why this was done are still unknown.
- Allowed Manufacturing changes and improvements over the 10 year period were likely done to improve the DCVax-L vaccine
- Less than 50% of the Methylated patients were 36 months from surgery from the original 182 (Methylated MOS = 34.7 months, as given by the paper). Considering the rate of 36 month OS is significantly better for the Methylated than the Unmethylated (46.4% vs. 11%) this should lead to an increase in ITT 36 month OS.
Let’s consider these numbers and take a look at what they mean:
- The first 182 showed a 25.4% 36 month OS
- The overall BLENDED ITT population is “expected” to be 28.1% for 36 month OS (from calculations above)
So now let’s project how many of these were the DCVax-L treatment arm vs. the control arm patients? We can estimate these figures with the following statistics:
- Historical data for the control group which is the current standard of care for treating GBM, projects the chances of surviving 36 months at 16%
- There were 99 patients on control, and 54 patients crossed over to receive DCVax-L once their cancer progressed (as noted, this was part of trial protocol)
- Due to the crossover, in my opinion, the control is likely to do better than 16%
- I am going to predict that the control will show a 15%-20% increase in 36 month overall survival than historical data (and this will be due to the crossover to treatment aspect of this trial)
- This would mean our control group will have a 19% chance of survival at 36 months (20% x 16= 3.2, 3.2 + 16 = 19.2% or 19%)
- According to our previous statistical calculations, I predict that 28.1% of the entire ITT group will reach 36 month overall survival
- This would mean that since the control group is expected to see a 19% survival at 36 months, we can estimate that there will be a 32% survival for the treatment group at 36 months
- These figures can be reached by using the following math:
- Treatment= 232/331 = 70.1% of the trial was made up of treatment
- Control= 99/331 = 29.9% of the trial was made up of control
- Powered for randomization-
- 70.1% x 32= 22.4
- 29.9% x 19= 5.7
- 22.4 + 5.7 = 28.1% blended 36 month OS (as predicted)
These numbers would conclude that the DCVax-L arm shows a 100% improvement over the historical standard of care of 16% 36 month OS, and a 68% improvement over the estimated 19% 36 month OS for this trial’s control group! Optune was approved with a 63% increase over the control group SOC, and was considered a “blockbuster", and major medical advancement. This estimation involves numerous calculations, statistics, and predictions based on observed attrition rates, and historical values. However, I am very willing to argue that the 32% OS for 36 months for the treatment group (DCVax-L arm), is a conservative prediction. I argue this is conservative due to a few key observations that are very likely to additionally improve the 36 month OS:
• I believe the “2nd part” patients in the trial will show a much improved 36 month OS and the figures I’ve cited thus far are based on the interim “1st part” patient data in the DCVax-L trial
• The 2nd part of the trial will have a greater ratio of patients in the treatment arm for DCVax-L
- 1st part (182 patients)
- Treatment = 122
- Control = 60
- 122/182 = 67% treatment in 1st part
- 2nd part (149 patients)
- Treatment = 110
- Control = 39
- 110/149= 74% treatment in 2nd part
• Over the last 10 years, it’s plausible that there were manufacturing optimizations/improvements made
• These estimations were based on attrition rates from the first part of the trial. The methylated and unmethylated attrition rates for the 2nd part of the trial are likely to improve over the attrition rates from the 1st part of the trial if there was any improvement in manufacturing after the trial transitioned to a Phase 3 trial, because the 1st part of the trial patients appear to date back to when the trial was a Phase 2 trial
Regardless of any manufacture improvements, if you simply take the estimated 32% OS for 36 months for the DCVax-L arm, and compare it to the predicted 19% 36 month OS for the control arm, you still see phenomenal data. This would represent a 68% improvement over control (32-19= 13, 13/19 = 68%). To add to this impressive data, side effects observed from the DCVax-L treatment were practically non-existent. As a clinical pharmacist I have seen first hand the effects chemotherapy has on patients. I can assure you an improvement in quality of life when on cancer therapy would be very welcomed by patients. While the Phase 3 trial did utilize the chemotherapy drug Temozolomide (Temodar), DCVax-L may help to mitigate the terrible side effects of chemotherapy, and perhaps in some cases, eliminate the need for it altogether.
Is it possible we may soon see a "blockbuster" therapy for GBM? It seems quite likely to me.
Disclosure: I am/we are long NWBO.
Additional disclosure: I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it. I have no business relationship with any company whose stock is mentioned in this article.