Robot assistance in the laboratory

Drugs, whether as tablets or other dosage forms, contain not only active pharmaceutical ingredients but also various additives to ensure that the active substances remain stable and that the intended dose is released at the right place in the body. These additives differ according to the drug. In order to find out which formulation best achieves the desired medicinal effect, thousands of tests are necessary. “It is not enough to produce different additive combinations,” explains Oliver Germershaus from the HLS Institute for Pharma Technology. “We also have to analyse those formulation candidates chemically and assess their shelf life, which means: storing them for long periods of time and then testing them. Many of these steps are still done manually; the process is vulnerable to errors and it takes a long time, especially for biological agents.” In order to automate the production of different formulations as well as the sample preparation, the HLS researchers have adapted a commercial robot platform for this special purpose in a project funded by the FHNW Foundation and in collaboration with Hamilton Bonaduz AG.

Glass not plastic

The biggest challenge of the project was the biological agents themselves: biomolecules degrade over time. They are less stable and are more sensitive than conventional active ingredients, for instance when they come into contact with air. “Conventional laboratory robots work with plastic vessels,” says Germershaus. “However, these are not completely airtight. If samples are stored for shelf life studies, air can get in and may affect drug stability. More- over, substances can leach out of the plastic and influence the efficacy of the drug.” HLS researchers modified the robot and replaced the plastic containers with glass ones, which do not react with the drug, are airtight and can be sealed, ensuring more reliable test results. The research team used a monoclonal antibody, i.e. an artificially produced biomolecule, as a test system. Different concentrations of six different additives, such as stabilisers and buffers to regulate acidity, were used for the formulation. A total of 324 different combinations were produced by the laboratory robot. “The robot has several advantages,” says Germershaus. “It is about one and a half times faster than humans, can produce the formulations very reliably and does exactly what it was programmed for. In addition, the robot has a clean-air hood to filter the air and remove particles and micro- organisms, enabling it to produce the formulations under sterile conditions.”

IT meets pharma technology

The research team did a great deal of the work on the computer. Due to the many possible test substance formulations, this programming was very time-consuming and only feasible with the help of researchers from the Institute for Medical Engineer- ing and Medical Informatics at the HLS. First, they had to adapt the robot and clearly define its new tasks for each step, such as: go to this location with the pipette, take one millilitre of the active substance solution, go to the next location and mix again. Germershaus believes that despite extensive process automation, the laboratory robot cannot completely replace humans: “Humans see an experiment with all their senses. A robot works systematically but not necessarily efficiently. A human notices this immediately, but an automatic mechanism only checks the things it has been programmed for.”

Chemical analysis also relies on that combination of human expertise and IT efficiency. “Using statistical test design, we were able to reduce the 324 formulations produced by the robot to just 40 in the analysis phase and still find an ideal combination,” says Germershaus. His team stored the 40 formulation candidates at two different temperatures and analysed their effectiveness at four different points in time, meaning that the researchers still analysed 320 different samples using a variety of methods. For this purpose, they developed a new software tool that provides a clear visual presentation of the test results. The researchers see potential for further auto- mation in future. The robot and its algorithms could take on sampling and analysis for example, as well as the interpretation of the analysis results. However, it will probably be some time before the entire formulation process is fully automated.