Skip to main content

Andreas Lindauer at the annual conference of the German Pharmaceutical Society in Saarbrücken

In recent years immunotherapy in oncology (aka: immuno-oncology) has become a game-changer in the treatment of cancer. Several therapeutic strategies to stimulate the patients’ own immune system in order to attack and eradicate malignant cells are under active development with remarkable results in the clinic. These strategies include cancer vaccines, oncolytic viruses, transfer of ex-vivo activated T-cells as well as antibodies and therapeutic proteins that block the immune check-point pathway [1]. Ipilimumab was the first immuno-oncologic antibody reaching the market in 2011, targeting the anti-cytotoxic T lymphocyte-associated protein 4 (CTLA-4). More recently, in 2014, the anti-programmed death (PD)-1-receptor antagonist pembrolizumab was approved for the treatment of advanced melanoma with remarkable response rates in general, and even achieved complete disappearance of tumor lesions in a number of patients [2]. Clearly these drugs bear an enormous potential and new antibodies targeting different immune check-points (e.g. LAG3, TIM3) as well as combinations of immune-therapeutics are being developed [1].

A quantitative understanding of the pharmacological mode of action as well as the pharmacokinetic behaviour of these antibodies is paramount to identify the most promising candidate compound or combination and to define the most appropriate dose for patients in clinical studies. It is desirable to have an understanding of the most likely efficacious dose level already early in the development process of monoclonal antibodies. At this stage the production of an antibody is usually done at a small scale and requires careful planning of resources. More importantly, however, Phase 1 studies in oncology are usually already conducted in patients and the dose given should have a reasonable chance of being efficacious.

Translational pharmacokinetic/pharmacodynamic (PKPD) models allow the prediction of a range of likely efficacious doses early in the development. Information on drug-specific aspects (e.g. receptor binding, tumor growth inhibition in mice) from in vitro and in vivo experiments can be combined with data from the literature on system-specific parameters (e.g. blood/lymph flow in mice and man, distribution volumes) to form a mathematical description of the pharmacological system based on the current state of knowledge. Once an adequate model is available to describe pre-clinical in vivo data, model parameters are translated from animal to human making assumptions on size-dependency of parameters and the degree of conservation of pharmacology in different species.

The author had the opportunity to work in a team developing a translational PKPD model for the anti-PD1 monoclonal antibody pembrolizumab which was published recently [3]. In the talk the translational modelling approach will be illustrated using pembrolizumab as an example. Albeit in this case some clinical data (PK) was already available at the time of modelling, efficacy studies in patients with metastatic melanoma were still ongoing and decisions around the minimally efficacious dose needed to be made before the clinical results became available. The predictions with the translational model filled the gap and supported the sparse clinical data available at the time of decision-making.

While the pembrolizumab example is a success story as the predicted efficacious dose range matched very well with actual clinical results in retrospect, there are certainly pitfalls and uncertainties in the translation of a complex biological system like cancer and the immune system [4]. The limitations of the approach will be discussed as well as strategies to handle uncertainty and assumptions inherent to the model. The shortcomings will be put into perspective with the great potential that translational modelling has to improve the development of new immuno-oncologic drugs.


Acknowledgments: My former colleagues at MSD who were involved in the development of the pembrolizumab model (C.R. Valiathan, K. Mehta, V. Sriram, R. de Greef, J. Elassaiss-Schaap and D.P. de Alwis).



  1. Farkona, S. et al.: BMC Medicine. 2016, 14:73
  2. Robert, C. et al.: N Engl J Med. 2015, 372, 2521-32
  3. Lindauer, A. et al.: CPT Pharmacometrics Syst. Pharmacol. 2017, 6, 11–20
  4. Stroh, M. et al. : CPT Pharmacometrics Syst. Pharmacol. 2014, 3, e128

How can we help you?