A biomarker model that is almost 100% predictive
Michael Cannarile and his colleagues discovered biomarkers that helped to identify the optimal dose for a new cancer drug. The use of these biomarkers for similar therapeutic molecules is gaining acceptance at Roche.
As a scientist, it is incredibly exciting to receive confirmation that an experimental molecule actually works in humans the way you hypothesized. That is especially true when the molecule was designed to fight a life-threatening disease like cancer.
I became interested in science as a teenager. I could not understand why there was no available cure for cancer and people had to die. I was determined to understand the biological basis of this disease. That is why I ultimately went into pharmaceutical research and joined Roche in 2009.
I work as Biomarker and Experimental Medicine Leader for Roche Pharmaceutical Research and Early Development in Penzberg, Germany. My focus is on developing biomarker strategies to explain and predict responses in patients treated with drug candidates. In simple terms, a “biomarker” is any biological parameter used as an indicator of disease process or drug response.
At Roche, biomarker teams play a vital supporting role that links Pharmaceuticals and Diagnostics. In the early stages of discovery, these biomarkers can provide us with insights on the disease biology and the mode of action of a therapeutic molecule.
Our Diagnostics colleagues are involved at the early phases of a project – a collaboration that is quite unique to Roche. Their ultimate goal is to develop a reliable biomarker-based test, known as a companion diagnostic, which will help us predict which patients will respond to our targeted therapies. This is central to our concept of Personalized Healthcare.
In 2010, we were nearing the end of pre-clinical testing of of an investigational antibody which binds to CSF-1 receptor expressing cells. This receptor is overexpressed on the surface of some tumour cells. The same receptor is also present on specific types of macrophages, white blood cells that are normally a beneficial part of the body’s immune system. These macrophages are present in many tumour types.
Tumours sometimes “hijack” these macrophages and use them to stimulate tumour growth and prevent other cells from attacking the tumour efficiently. Therefore, colleagues in Drug Discovery developed an antibody that binds to and eliminates CSF-1 receptor expressing cells.
Up to that point, experiments with this investigational antibody had only been performed in laboratory pre-clinical tests. We still did not have data with the clinical version of the antibody targeting human macrophages.
The next step in the clinical development of the antibody, per regulatory requirements, was to test its safety profile. Here’s where we saw an opportunity. Beyond safety parameters, we wanted to know whether our biomarker hypothesis and assays were accurate and whether there was a dose-dependent response to the therapy. Another goal was to develop a mathematical model correlating the drug exposure to the biochemical and physiological response of the body.
Amending the standard toxicity testing protocol, however, presented some challenges. The implementation of biomarkers increased the complexity and the budget significantly. Furthermore, the additional data generation and interpretation represented a potential risk for the tight development timelines for Phase I clinical testing. Hence, we needed to explain the expected benefits for the patient to our colleagues. In the end, with the help of colleagues from Discovery and Pharmaceutical Sciences, we successfully incorporated these markers into the study protocol.
The results were very encouraging. They supported our hypothesis on the mode of action of the antibody, validated our assays and provided a rationale for the appropriate dose to treat the first patient. That meant we did not need to expose study participants to insufficient or excessive doses of the therapy, and thus potential side effects. By shortening standard dose escalation procedures, we also accelerated the timelines of the Phase I trial.
Using biomarkers early enabled us to avoid exposing study participants to excessive doses of the compound, and related potential side effects.
Now that we have progressed in phase I monotherapy studies with this new cancer drug candidate, we can look back and see that our biomarker model generated in preclinical testing was almost 100% predictive of what happened later in humans.
As a result, the implementation of biomarkers for pharmacodynamics (biochemical and physiological responses) as well as mode of action in primate studies is gaining more acceptance at Roche for this new type of targeted therapy. Further clinical development of this antibody will focus on identifying a predictive marker that enables physicians to select patients with the highest likelihood of response. The identification of a predictive marker will trigger the development of a companion test for the safe and effective use of the therapeutic drug.
To me, this is what Personalised Healthcare is all about – knowing which medicines will work for a specific group of patients. We’re not there yet, but it’s an important step in understanding cancer and how best to treat it.