Definiens shines a light on growing importance of biomarkers


As scientific understanding of oncology’s complexity grows, and therapies become more and more specific, it is vital for the key players to try to identify biomarkers to pinpoint relevant groups of patients and guide development of new therapies.

But beyond starting studies to identify such markers, or signing deals with companies that specialise in screening assays, industry work has tended to shun the limelight. This makes the $150m bet AstraZeneca made yesterday to acquire the private German group Definiens particularly striking.

Of course, the burgeoning field of immuno-oncology has complicated matters. Things are no longer as simple as identifying a particular tumour subtype, say Braf-mutated melanoma, and developing a specifically targeted agent against it.

This, Astra claims, is where Definiens comes in. The German company has developed a computer software-based technology that uses proprietary mathematical algorithms to investigate the interaction of immunocytes with tumour cells, says Ed Bradley, head of the innovative medicines group at Astra’s Medimmune R&D unit.

“For each patient you’re looking at millions of correlations, and if you process data across small subsets of patients you’re making billions of correlations,” he tells EP Vantage. “This goes far beyond what any individual pathologist – looking at one slide, analysing one marker like PD-L1 – is able to do.”

Changing microenvironment

This is just as well, since it is abundantly clear that the tumour microenvironment in which checkpoint inhibitors operate is constantly changing, and certain cell surface targets are only expressed after an initial line of treatment. Thus there is limited correlation between patients’ PD-L1 status and their response to PD-1 blockade, making PD-L1 a pretty poor biomarker.

On this basis oncologists are leaning towards the view that PD-L1-negative patients should not be deprived of anti-PD-L1 or anti-PD-1 therapy, preferring a more general prognostic marker such as infiltration of T-cells into tumours (Asco – Growing PD-1 uses fail to stem Bristol selloff, June 3, 2014). A lack of standardisation among the immunohistochemistry assays being used is a separate problem.

Not that this has stopped companies looking for a correlation. Merck KGaA’s head of the immuno-oncology, Helen Sabzevari, recently told EP Vantage that her company had developed a diagnostic kit to test for PD-L1 status, for potential use with its agent MSB0010718C.

Just two weeks ago Clinical Cancer Research reported that tumour PD-L1 mRNA expression was associated with improved outcome in breast cancer, and joined the dots, correlating this expression with increases in tumour-infiltrating lymphocytes.

Mr Bradley says the Definiens system is able, in any tumour specimen in any setting, to identify all the immunocytes involved in the tumour response, characterising the cells in great detail. It also identifies which tumour cells are growing and which are being killed, as well as pinpointing the immune regulatory cells at the site of the tumour.

“Then it examines each of these relationships but with millions of data points, even billions of data points, in an individual patient, and in an agnostic fashion identifies the correlation. It’s a powerful technology that ... sorts through the very complex process whereby multiple cellular actors play a role in either enhancing or blocking the immune response to cancer.”

For Astra there are many benefits. Understanding the biology will inform the sequence of therapy and the types of combinations that can be used, helping design studies and perhaps speed up development. And after treatment it could help figure out patients’ prognosis.

“Rather than just administering a therapy and waiting many months [for] large numbers of patients, we'll look at every individual patient before and after treatment and see exactly what happened immunologically, what cells were important and what effect our drug had,” says Mr Bradley.

A cure... for some

A key driver of individualising treatment is that checkpoint inhibitors are known to work only in subsets of patients, but in such responders their effect is striking and long-lived. While targeting patient subpopulations will reduce each available market, companies might look to charge premium prices for offering “cures”.

Clearly all the big immuno-oncology players are active in this area. Bristol-Myers Squibb has run biomarker studies, including with Yervoy, and last year struck a biomarker collaboration with Adaptive Biotechnologies. Roche and Merck & Co list biomarkers as high on the list of development priorities for pembrolizumab and MPDL3280A.

Astra’s $150m is quite a bet to make, but at least the group knows what it is buying, having been a technology client of Definiens for almost 10 years. Mr Bradley says the technology is unique, adding: “I’m not aware that any other pharma company has exactly this within its shop.”

He says Astra will look to share the Definiens technology with others, while internally it will use it not only for immuno-oncology, but also for targeted and small-molecule approaches, and in any setting where immune response needs to be characterised.

“When you see additional combinations going into the clinic you can be sure that Definiens data went into that to guide us,” says Mr Bradley.

To contact the writer of this story email Jacob Plieth in London at or follow @JacobEPVantage on Twitter

Share This Article