Thx for this work. I am not the scientist input you are seeking, but:
- if you start with the groupings looking something like neural = 10%; proneural = 20%; classical = 35%; mesenchymal = 35%,
and then assume the order of effectivity strength of SOC is approx. mesenchymal at the bottom followed by classical by neural by proneural at the top,
and then assume 25% of the 35% of the mesenchymal pts flip from the bottom to the top priority, there is the potential of a nice change in the medians of both PFS and OS if you can move the median out of the SOC treatment bottom dwelling sets of mesenchymal and then classical and progress into the neural sub-type when determining DCVax-L therapy medians. This is why FDA wants to see a biomarker to understand who is responding and who is benefitting.
I think the complicating factor in understanding sub-type correlations in GBM is the heterogeneity of the tumors, which seems to show that there are combinations of phenotype sub-types in each tumor, which then means the accuracy of any test needs extensive work to validate and rules of composition to make clear determinations have to be developed,... there is more but I will hold up, such as speed and cause of mutation and whether certain sub-types are dominant in behavior,... I wonder about the accuracy of the early tests on pt tumors and how dependable those determinations were as knowledge grew about the heterogeneity of GBM tumors. It does not appear to be so black and white until more work is done to track tumor composition and the changes to that composition over time and over time in relation to therapy, which can change the course of the mutation. I don't think the more accurate single point in time tumor composition tests that are discussed now were always in existence nor in general use. Maybe UCLA was involved in the test development but I don't put too much weight on the accuracy of the body of knowledge of GBM tumor compositions to date, but, rather it is directionally usable. I would like to see more data on the complete composition of some tumors. That would take the saving of complete resected tumor tissue and testing of all the tissue as much as possible throughout the entire tumor, not just using small tissue samples. This is a great example of why work such as this can take decades and not just years. Complications of tissue sample availability, test method development, new learnings about the test capabilities given the samples are live tissue,... All this just to understand what one is trying to solve, then on to designing, developing and proving the solution. It is easy to see why all the steps have to be conducted concurrently. And, easy to see clinical trials are not as easy as performing tests on pts and collecting results.
I agree with you this is an area of collaboration with FDA by NWBO and a necessary area of work particularly if an immune response is seen in a limited number of pts. If a good to a very good response is seen in these pts the work is absolutely worthwhile to try to ring fence the correct pt population. If there are little safety concerns involved the ring fence can be somewhat broader than if there were safety concerns. Generally speaking if a treatment does not help everyone or at least elicit a positive response in most pts the FDA will want to know who it helps so all pts are not subject to it with a good percentage having little chance of being helped. This is considered more heavily when a treatment is additive to SOC and not an alternative treatment. But, keep in mind DCVax-L has virtually no negative QOL effects, another reason why I think FDA would tend to be very careful about restricting the treatment too tightly. And, there may be other longer term benefits from early treatment with DCVax-L, TBD shortly.
So, given the relative inaccuracy that would be possible and necessary for such testing for a DCVax-L pt population and the potentially significant benefit that could result for the pts a relatively accurate testing would probably suffice. And, the testing may not restrict the application of the therapy to a subset of pts but rather serve to continually grow the field of knowledge of tumor composition and their respective response to the therapy. The use of a biomarker as of a point in time may suffice for treatment applicability decisions, assuming it accurately reflects at least the general or majority tumor composition. When one pauses to consider the depth and breadth of what is needed to present to FDA in a BLA, it is quite remarkable such a small company can undertake the work. This is more the concern I have than the effectivity of the treatment. FDA will want a deep understanding of the treatment's MOA and other effects in addition to its effectivity vs GBM tumors. It can lead to significant gains in knowledge for treatment evolution. Once a treatment has been approved the use of it to learn is geometric given the chance to collect so much more data. BP's are so much better equipped to perform all development work and to capitalize post approval. Folks who think a company such as NWBO can do it alone are just not privy to what occurs behind the industry. Since DCVax-L has demonstrated little safety and QOL effects to worry about I am hopeful the benefits demonstrated will carry the day until NWBO partners with a BP soon.
AI
