Cancer can’t be addressed in isolation with any one approach because by nature it is highly heterogeneous and continually evolving. This isn’t news to anyone working in the field yet the very nature of academic discovery and translation leads to cancer therapeutics being developed to address one isolated aspect.
The probability of surviving nearly all types of cancer has increased over the last 10 years currently ranging from 11% in lung cancer (depending on stage of detection) to 90% in testicular cancer. In those cancers with lower survival rates therapeutic interventions fail for what are largely foreseeable reasons. Either whichever marker or metabolic profile they are targeting occurs elsewhere, so they are toxic beyond tolerable levels (off-target effects). Or they fail to reach every cancerous cell. Or they reach a toxic level, thereby asserting selective pressure which drives resistance. Or simply fail to take into account a large component of a complex system, particularly when it comes to the tumor microenvironment and the action of the immune system on delivery vehicles.
Biology and engineering have moved far beyond the days when ‘science is hard’ was a reasonable excuse, however the process of translation, venture and clinical practice hasn’t caught up and often still operates in silos: even where multi-disciplinary work is essential. Under this paradigm, it will take a long time to tackle the complexity of cancer. Our hope is that with our network driven approach and partnership with the world’s largest independent funder of cancer research, Cancer Research UK, that we can take a step in the right direction. Over the last year we’ve completed our ‘near term’ strategy (see below) and now we are looking at how we can tackle the ‘long term’ challenges.
Near term: Improving the effectiveness and applicability of immunotherapy.
In the near-term we are looking at how we can increase the effectiveness and applicability of recent breakthroughs in immunotherapy. To this end, we have recently launched two new ventures ConcR (reducing the uncertainty in treating cancer) and ImmTune (removing manufacturing from autologous cell therapy). We are also investigating whether there are any gaps in early detection and how insurance or the NHS can better accommodate highly customised therapies: if either of these are areas that you are passionate about, drop us a line!
Long term: Equipping therapeutics to tackle the complexity of cancer
With a longer term view we are looking at how can we finally tackle the complexity of cancer. We’re starting with three areas which we believe largely covers the gamut of the problems space: 1) focusing on equipping therapeutics to act differentially and specifically to the local tumor-microenvironment, 2) equipping therapeutics to provide 100% coverage with one dose, preventing escape and resistance without relying on a long tail of combination cocktails, 3) addressing oncogenesis by designing synthetic lethality that can act as a more general sensor to genetic errors than needing to be specifically discovered and encoded into the therapeutic.
Equipping therapeutics to tackle the complexity of cancer
Enabling immunotherapies in the solid tumour microenvironment
Immunotherapies have demonstrated incredible results in blood-based cancers and have increased patient survival in breast cancer and a limited set of other solid tumour but by limited degrees, often only 10-20% vs. existing therapies. Recent trials have shown that antibody-based checkpoint inhibition can significantly improve outcomes. However, this solution alone doesn’t work in all cancers, or all patients, and in addition, places selective pressure on the cancer to evade this method of detection, leading to resistance and recurrence.
The incentivisation structure of R&D leads to the bulk of focus being on finding the next target (TIM3, LAG3, TIGIT, VISTA or one of many cytokines and other up / down regulating messengers), running another trial with the best in class immuno-therapy at the time, sometimes stratified appropriately, but often not, and often leading to the same result. Either it works in a subset of patients briefly but ultimately leads to resistance or fails completely. Over time clinicians will work out which combinations work best, but we cannot help but question whether this linear approach is really the best way of exploring a highly complex and evolving space.
How could we move the focus away from blocking singular targets and ‘n of one’ stratification of trials? Could we instead create therapeutics that learn and adapt with their environment? Could we wipe out both the microenvironment and cancerous cells where a sufficient number of signals indicate a diseased state? How could we ignore useful inflammation? How could this, and further management of complexity be integrated into one therapeutic?
The 4th gen CAR-Ts are a step in this direction, utilising several different methods, including logic gates, to increase specificity and release of cytokines. Increased specificity can also drive escape, however, so the 5th generation is likely to involve a much more subtle, differential environmental response more akin to that seen in microbiome therapeutics (including our own portfolio company, CC Bio). How could this be extended to the immune system leveraging the broader synbio toolkit and research in the dynamics of complex systems?
Full identification, full clearance
Relying on surface, or even internal markers will always be a flawed strategy because they are never universal, constantly changing, and are often completely hidden during intermediate states, creating an advantage for the genetic profile not caught by the initial therapeutic dose.
One approach that has fallen in and out of favour over the years, is to use viruses which can selectively infect cancerous cells due to specific specific genetic or metabolic phenotypes, such as unregulated replication. This can then be used to lyse cancer cells, display a marker to the immune system, produce cytokines or prodrug convertase. The problem with this approach is that the virus are highly unlikely to make it into all of the cancerous cells because it is quickly spotted by the immune system and destroyed. Either pre-existing or adaptive immunity presents a problem, the immune system can destroy the virus quicker than it can destroy the tumour. The result is again being impressive initial results followed later by what is often even more aggressive recurrence.
Our starting point to explore this space is to ask how we can create a more effective balance between, on the one hand, evading the immune system and on the other hand, getting into every cancerous cell whilst pervading the tumour microenvironment. Next, once we have achieved this selective penetration, how do we drive a therapeutic regime that does not allow resistance to emerge?
Next generation synthetic lethality
Most oncogenic alterations in humans (mutations, translocations, amplifications, deletions, and epigenetic modifications) are caused by the inefficient repair of damaged DNA. These errors occur at a rate of around 10,000 a day mainly due to genotoxic insults from endogenous reactive oxygen species (ROS). This is a disease of aging, it is only a matter of time before cellular repair machinery fails to catch an error, or more typically miss-repairs it, and this leads to the development of a population of cells with oncogenic potential.
One of the highest potential avenues of investigation in oncology at the moment is targeting deficiencies in cellular DNA repair machinery that are unique to cancerous cells. Tumor cell deficiency in tumor suppressor gene p53 has long been a target to increase sensitivity to DNA damaging therapeutics ranging from cisplatin to radiotherapy. This is because, by comparison, normal cells are better able to repair, although unfortunately this still drives damage and further tumors.
More recently ‘synthetic lethality’ has emerged as a more advanced method of leveraging cancer’s greatest strength; the lack of checks on DNA, against it. This adds a layer of selectivity (or logical computation) to therapeutics by targeting non-druggable cancer-promoting lesions with pharmacological inhibition of the druggable synthetic lethal interactor. In theory this should select exclusively cancer cells, and be well tolerated by healthy normal cells, that lack the cancer-specific mutation. Impressive clinical results have already been achieved with PARP inhibitors (PARP enzyme repairs double stranded breaks) in patients with BRCA1/2-mutated cancers (BRCA repairs double stranded breaks), including ovarian, breast and prostate cancers.
The ability to add this level of selectivity in what to date has been a brute force approach is really exciting but it’s still far from perfect. The best results so far are in breast cancer, where patients that have this mutation had a 33% response rate, much better than other therapeutics, but 67% still died and the long-term outcome remains to be seen.
Our concern is that we are about to start another long journey of identifying targets, in which it might work in 1/3rd of patients, whilst the very nature of the genomic instability and population heterogeneity of cancer means that there is probably a way around any single point weakness. We are starting with the question of how can we detect abnormalities in DNA, mediators and effectors, not by sequential target discovery and development, but by equipping the therapeutic with the inherent ability to discriminate between overall normal and states which have oncogenic potential.
Clearly all of these are moon-shot projects in themselves, with thousands of people already working in these areas and billions of pounds in research funding. We hope that our approach of bringing together diverse fields, and often asking the non-obvious questions, can create new directions that might play some small part to accelerate the goal of eradicating cancer.