Case Studies

Automated Salts and Polymorph Screening

Salt and polymorph screening is an important step in the drug development process and consists of an investigation of various physical properties of the drug substance. The aim is to identify and evaluate salts of the drug substance and then to identify the different crystalline structures (polymorphs) that the drug can adopt. This information is used to select the best physical properties for the drug substance which will ensure efficacy and provide consistency during formulation and manufacture.

GlaxoSmithKline initiated an internal project to improve their screening process to make use of the latest technologies in automation and solid-state analysis. The aim was to significantly increase the range of experiments that were performed per compound, and to meet the demand to screen a higher number of compounds earlier in development.

Tessella worked with a core team of users to analyze the new workflow to develop a software system that would control the automated equipment, as well as providing workflow coordination, data storage, sample tracking and a web based system that allows customers better visibility of the results. Tessella then designed and developed a solution that, used in conjunction with hardware developed at GSK, is delivering improvements in data quality and increased throughput, and providing a framework in which to research and develop new processes.

The software was released in phases to both the UK and US, each with increased functionality, which gave the users an opportunity to use the system at an early stage and to provide feedback into the development process.

Much of Tessella’s work during 2004-2005 has concentrated on enhancing the system to increase the efficiency of the workflow, and incorporate the latest developments.


Once a drug substance has been discovered, it is necessary to consider how it will be administered to the patient. A drug substance will often have inappropriate physical properties, such as poor solubility in water, a lack of crystalline form, or poor stability. These problems are often addressed by creating a ‘salt’ version of the drug substance, in which a basic or acidic drug molecule is paired with a ‘counter-ion’. Pharmaceutical companies frequently carry out a salt screen, which is the process of evaluating potential counter-ions and the properties of the resulting salts, with the aim of identifying a salt version that is suitable for manufacture and formulation.

Materials that are crystalline in nature can often adopt a number of different crystalline forms, or ‘polymorphs’. Different polymorphs will have different physical properties (e.g. solubility, stability) which can often impact the manufacturing process and sometimes the bioavailability, and hence efficacy, of the drug. The regulatory authorities expect pharmaceutical manufacturers to have conducted a thorough assessment of the propensity of new drugs to form polymorphs and to have implemented appropriate controls within the manufacturing process.

The exact polymorphic form adopted in a given situation depends intimately on the crystallization conditions. It is therefore very important for a pharmaceutical company to understand the crystallization conditions, so that the required polymorph can be manufactured reliably. If the manufacturing process suddenly started producing a different polymorphic form, particularly if the physical properties of the new form meant that it could not be formulated for dispensing in the same way or equivalent bioavailability could not be achieved, then the drug may have to be withdrawn form the market.

Another serious issue for pharmaceutical companies is that it is common to find metastable polymorphs that revert to a more stable form over the course of time. This can have serious implications for the shelf life of the product and so it is necessary to investigate the long-term stability of each identified polymorphic form.

Pharmaceutical companies address the issues associated with polymorphism by carrying out a polymorph screen on the chosen compound, be it a salt version or free acid, free base or neutral molecule. The aim of this process is to identify as many different polymorphic forms as possible, to understand the conditions under which each polymorph is produced, and to profile the stability of each polymorph over the long term.

Ideally, salt and polymorph screening will take place at an early stage in the drug development process, so that a manufacturing process can be defined to produce the right form for clinical trials and ultimately, the market. If a form switch is made later in development then expensive and timeconsuming bio-equivalence studies may be required. However, the screening process consumes time and resources, and there is no guarantee that all polymorphs will be discovered. It can also require a significant amount of API, which may not be available during the early stages of development. It is therefore essential that screening and selection of the desired salt and polymorph is carried out efficiently.

Business Problem

Traditionally across the pharmaceutical industry, salt and polymorph studies were carried out manually by the various individual chemists who were working on a particular drug substance. This approach was inefficient and did not make best use of specialized know-how. Each chemist employed a slightly different approach, and this lack of consistency meant that it was difficult to understand and apply the results elsewhere within the organization. It was recognized that there would be significant benefits from centralizing the salt and polymorph screening.

GSK provides an internal centralized salt and polymorph screening and evaluation service, which has two main goals, corresponding to key milestones in the selection of the preferred crystalline form for development:

  • To identify the preferred salt and complete an initial polymorph screen so that a suitable form can be used for 28-day safety studies
  • To select the final form so that optimum crystallization and characterization conditions can be defined in time to supply materials for Phase 2B studies

When this service was launched, the entire process was carried out by hand, meaning that throughput was limited. A simple automation solution was introduced to address the need to carry out a wider range of experiments, but following an industry-wide shift to push more compounds through the early stages of development, it became clear that the service would be placed on the critical path for some drug development projects.

The process of salt and polymorph screening generates a large amount of raw and processed data. It was possible to manage this information for the small number of compounds screened under the original manual process, but as the service grew the volume of data became a significant problem. This is because the data were distributed over a wide range of analytical machines and held in a series of individual spreadsheets on various file stores. The long-term nature of the screens made it even harder to keep track of this information.

Project Execution

Tessella worked with many key personnel within GlaxoSmithKline to automate the salt and polymorph screening process. Staff from the Applied Technology Group and the Chemical Development Strategic Technologies department designed and implemented a hardware platform, consisting of modified solid and liquid handling robots and a range of custom-built reactors and sample vessels. Tessella implemented the software to design and execute experiments using this hardware, and to assist with the analysis and reporting of results.

Tessella worked with a core team of users to analyze the screening process and to establish the requirements for the software. A series of phased releases, each with increased functionality, gave the users an opportunity to use the system at an early stage, and to feedback comments into the development process. Tessella then provided onsite support for the rollout of the system.

System Overview

When a chemist requests a salt or polymorph screen, a user of the system will enter the details of the request, including key information about the batches of sample supplied. The user will then design the screen. The first design step is to select the main protocol (or recipe) that the screen will follow. The protocol consists of the initial plan of operations that are to be carried out for each experiment in the screen. Once the screen has started, the user can edit the protocol in order to accommodate any changes that are required (e.g. it may be necessary to repeat a series of steps under slightly different conditions if they didn’t work the first time). The screening process will inevitably change over time, so the users can define new protocols for use at design time.

Having chosen a protocol, the next step is to specify which experiments will be carried out. For a salt screen, the aim is to evaluate potential salt versions of the API, so the user will pick a selection of counter-ions. For a polymorph screen, the aim is to identify as many different polymorphic forms as possible and to understand their formation conditions, so the user will pick a selection of crystallization protocols (the input sample for these protocols is a saturated solution that is generated by the main protocol for the screen). The user will also select the different solvents in which the experiments will be carried out. The result is a two dimensional array (or table) of experiments, where the columns denote solvents and the rows denote either counter-ions or crystallization protocols, depending on the type of screen. A typical screen consists of between 50 and 400 experiments.

Once the user has defined the array of experiments that constitute the screen, the system calculates initial dispensing amounts and conditions, based on information such as material purities and solubilities that has been supplied by the user. If required, the user can edit the parameters for each step of the protocol on a per-experiment basis.

The array of experiments is a useful concept when designing the screen, and is also extremely helpful when visualizing, analyzing and reporting results. However, the actual experiments are carried out on plates, so it is necessary to transform from the array-based view of the screen to a platebased view. The system supports a wide variety of plates, which have a range of purposes and capacities (wells per plate, volume of each well). The user can also add new plate types as they are developed. When the user has finished designing the screen, the system automatically assigns samples to the most appropriate plate types, and positions samples in order to take advantage of various liquid handling optimizations. As with all stages of the software, the users can override the default and adjust the plate layouts to meet their specific needs.

As soon as the experiments in a screen have been assigned to plates, a list of tasks is generated for each plate, corresponding to the underlying protocols for the experiments on that plate. There are many different types of task, each of which has a customized workflow. All workflows, however, share the following common structure:

  1. Assemble the materials and plates required for the task
  2. Specify additional information that was not available at design time
  3. Perform the task (manual or automated, depending on the type of task)
  4. Enter results (this is transparent for automated tasks)
  5. Update sample states and enter comments

A task execution wizard leads the user through the workflow for each task, prompting for information where required and providing feedback on the progress of automated tasks. Materials and plates are identified and tracked using barcodes, and the system enforces any time constraints that are defined in the protocols. Results are stored in a central database for subsequent analysis and reporting.

Most of the sample preparation tasks have been automated, including liquid addition/mixing, solid dispensing, heating/cooling, shaking, and sample transfer. The automation platform controlled by the system consists of a mixture of off-the-shelf hardware (customized to a greater or lesser degree), and entirely custom-built hardware. All automated tasks can be carried out manually if required.

A wide range of analytical data is acquired during the course of most experiments, including Raman spectra, X-Ray Powder Diffraction (XRPD) patterns and optical microscopy images. GSK use a technique developed in-house to process this analytical data and identify the various end products that have been formed by each experiment (e.g. remnants of the starting materials and a variety of different polymorphic forms). The system presents the experiments as an array, and the user can enter the results of the analysis for each experiment. This array view enables the user to identify and explore trends across groups of experiments – this would not be possible with a plate-based view of the results. There are also tools for highlighting groups of experiments and for comparing the findings obtained from different, but complementary, analytical techniques. The final result of the analysis for each screen, together with representative supporting raw data, is published for viewing by the chemist who requested the screen.

A web-based reporting system provides the users of the system with a wide variety of reporting elements which can be used to construct interim and final reports for the requesting chemist. The same reporting system enables customers of the service to view the progress of their screen, together with any published results.

Benefits to GlaxoSmithKline

The increased use of automation – both in terms of newly automated processes, and the simplification and streamlining of existing automated processes – has allowed GlaxoSmithKline to realize many benefits. In particular, it is now possible to work with smaller amounts of material, more accurately and reliably. This means that larger screens can be carried out, earlier in the development process, leading to selection of the correct salt version and sufficient understanding of its polymorphism to avoid costly changes later on. The increased automation also reduces the number of tedious manual tasks (and eliminates the errors that inevitably creep in during such tasks), leaving the scientists free to concentrate on other work. Crosscontamination is minimized by the use of automated sample transfer processes for both solid and liquid samples.

The introduction of a workflow system has also provided many benefits. The overhead of managing the screening data has been reduced, so users can work more efficiently and with larger numbers of samples, leading to an increase in throughput for the screening service. Users no longer have to keep a manual record of the current progress of each screen, and the state and location of each sample is managed by the system. All of the results are stored against the relevant sample in a central database, rather than being scattered around in lab notebooks and on various lab PCs and spreadsheets. This greatly simplifies the retrieval of results for reporting and analysis, and users can access the data via the corporate intranet. This is particularly beneficial for long-term studies, because it eliminates the need to retrieve handwritten lab notebooks from the hard copy archive. The new analysis tools enable users to identify trends within and across their screens in ways that were not possible before the results were stored centrally.

The combination of bespoke software and hardware enables GSK to provide an efficient, high quality screening service that meets the needs of a changing portfolio. The centralization and automation of the service means that a more rigorous and consistent scientific process is applied, leading to higher quality information that can be shared and applied across the organization. There has also been a significant increase in the number of compounds screened and the complexity of experiments, with no increase in equivalent resource. In turn, GSK benefits from significant cost savings and reduced time to market.