Driving lab automation forward

A round table discussion covering the driving forces behind the integration of automated technology within the pharmaceutical industry, the procedures that are followed when implementing new automated techniques, current areas of drug discovery most benefiting from lab automation, how lab automation advanced the drug discovery marketplace over the last five years, the groundbreaking techniques occurring in lab automation, the relationship between vendor companies and their customers and the current limitations within lab automation.

Question 1: What would you say are the main driving forces behind the integration of automated technology within the pharmaceutical industry?

Allenby: Due to the ever-increasing number of compounds in each Pharma collection, the internal pressures to decrease costs per well screened have become a major driving force. From a management perspective, in a fixed head count environment, implementation of automation can improve the number of compounds screened per target per FTE*, per annum. This is a tangible, if simplistic, measure of productivity. In addition, implementing automation can result in a re-alignment and re-utilisation of highly skilled resource: we are biologists not automation experts. I would argue, however, that unreliable automation is more resource demanding than manual screening; turning biologists into lab automation trainees. From a screening perspective, improved data fidelity is a key driver for lab automation.

Alton: The main forces for integration of automation today are attempts to lower costs via higher throughput. All companies are actively attriting research staff to lower R and D costs. Therefore, productivity per employee must increase to provide a stable drug pipeline. Automation allows for many more tests to be performed per employee.

Benn: In a broad sense, the driving forces for automated technology are the same as most other industry requirements for automation; namely that is throughput, consistency and capability. The first two indicators are self-explanatory, the third is basically pointing to the ability of a machine to do something that a human being could never do – picolitre volumes of dispensing is one major factor. However, this all comes down to the primary goal of increasing productivity within R&D.

Fransson: For the drug development and manufacturing QC areas which I represent, validated integrated GMP procedures eliminate the need for manual controls, such as data checking and result verification. A comprehensive automated procedure encompassing “request for analysis” for example, to final data extraction and reporting can yield significant savings. Such procedures also improve SHE and overall productivity.

Keller: In my opinion several factors contribute to and influence the decision to integrate automation technology. Firstly, there is the drive for ever more cost-effective and efficient research methods.

Secondly, there are the stringent regulatory demands on the production of medicine. For example, the PAT (process analytical technology) concept requires improved process and product understanding prior data submission to the FDA. This in turn, has lead to several initiatives in order to comply, such as the integration of more sophisticated analytical tools/instruments either on-line or off-line to monitor, if possible in real time, a number of chemical and physical properties simultaneously as well as the semi-automation of process steps.

Thirdly, a significant new driver is the “Quality by Design” initiative to regulatory submission, which will require much more experimental data to develop the required knowledge for the regulatory filing.

Fourthly, increased distribution of pharmaceutical products on a global basis brings with it the necessity to better understand and improve the collaborative nature of GSK’s interaction with its customers as well as to thwart counterfeiting.

Fifthly, there is the shift from batch to continuous processing and finally, safety, such as automated sample preparation which will reduce the exposure of analysts to highly potent drug products.

Question 2: Can you briefly describe the procedures that are followed when implementing new automated techniques?

Allenby: These procedures are; building, implement and testing.

Building

This begins with a user defined specification document (UDS) written in consultation with the scientists, followed by tendering, selection of supplier, close collaboration with the supplier to fully understand the UDS, constant project management and finally assay protocols are performed on the automation at the vendor site, followed by factory and site acceptance testing.

Implementation

Implementation involves the buy-in of the users, hardware and software training, plus designating a super user who should take ownership of the system: In my experience this should not be the project manager.

Testing

Testing involves using the equipment to perform assays and screening. Assays are modelled off-line using a single plate, moving it manually between the equipment. If, based on data quality, this passes quality control (QC), and then we write a protocol for the robot to repeat the process with a single plate. Once this is achieved, we screen a validation set of approximately ten thousand compounds run in duplicate on separate days to compare inter- and intra- assay variation. If the Z’ factors are above 0.5 with minimal automated QC then we will begin screening.

Alton: When a company implements new techniques for automation, the primary research group will evaluate the ability of the automation to enhance productivity. An investigation of return on investment (for example, value) is inherent. Performance benchmarks must be equivalent to the current standard protocol.

Benn: When implementing a new process, the first step to take is to thoroughly understand the existing process. I would recommend all team members, regardless of function or seniority, actually physically performing steps of the process which will be automated.

Once this has been performed, it is then often necessary to analyse the process. I repeatedly observe users exactly automating the steps taken in a physical process; this is not always the best approach. I would propose that the process itself is analysed at this point to determine efficiency savings and to determine steps which can be altered to be more ‘automation friendly’.

Finally, when you purchase – if you have any doubts over vendor specified functionality (and often even if you do not) then place acceptance tests in the purchase of the equipment contract and link this with staged payments.
Following these steps will, in my belief, assist in ensuring a successful purchase and implementation.

Fransson: Before implementing a GMP system, its compliance with the thought/work routine and the overall benefit should be critically assessed. Having done that, it is important that the implementation is defined as a project and has a sponsor. Validation should be carried out as part of the project, involving QA, and be mainly based on verifications made by an audited supplier. Apart from the user requirement and technical specification work, other aspects of the implementation such as facility changes, SHE/risk analyses, SOPs, training and support should be properly planned for.

Keller: Firstly, it is important to agree and manage business sponsor(s) expectations including deliverables. In addition to this, a full work flow of the process to be automated is a pre-requisite.

This is followed by Proof of Concept / Proof of Principle – generation of ‘real data’ to demonstrate functionality and utility of a technology, along with standard project management approaches: IQ / OQ / PQ or FAT / SAT. Following project closure there will be an after action review and an examination of business impact.

Question 3: What are the current areas of drug discovery most benefiting from lab automation?

Allenby: Cell-based HTS* has improved significantly with the use of lab automation. Fully automated cell-based GPCR* screening using FLIPR* has revolutionised this approach. The implementation of automated cell culture systems from T.A.P. and others has removed the “mystique” of cell culture. Recent innovations using methodologies employing frozen cells- to plate- to screen have improved assay simplicity and data fidelity due to more homogenous cell preparations.

Alton: The HTS market has been saturated for several years, therefore most of the new benefits of lab automation are occurring in high content screening which is most suited to secondary assays. These drive preclinical R and D programs following large HTS campaigns. In addition, various automated systems for protein crystallisation are becoming commonplace.

Benn: The traditional areas utilising laboratory automation such as high-throughput screening, sample storage and sequencing have benefited from laboratory automation, and this improvement is growing. In addition, the advance of benchtop automation and (albeit slowly) dropping prices have enabled other areas in the drug discovery process to successfully utilise automation. These areas, such as systems biology and ADME/Tox have seen large improvements due to laboratory automation acceptance in recent years.

Keller: The current areas of drug discovery most benefiting from lab automation include; large scale – S&CP (HTS screening) Compound Management (Sample storage and handling) and small to medium scale – some level of automation can be found in laboratories across the whole of the drug discovery process.

Question 4: How has lab automation advanced the drug discovery marketplace over the last five years?

Allenby: It is too early to say if lab automation has impacted directly on timelines to clinic. Compound management has probably witnessed the greatest process revolution over the last five years. The curation and management of a growing compound collection is not a trivial matter, nor is the delivery of millions of compounds in plates for screening in a timely manner. In addition, the capabilities within HTS to perform complex multi-step assays have been enhanced by lab automation, particularly in the area of cell-based screening. However, I would argue that lab automation has streamlined many processes within drug discovery through serendipity rather than design. Too many of us achieve our numbers objectives only to see our output fail later on in the pipeline. Quality and quantity are important.

Alton: Over the past 5 years scientific conferences have revealed that biopharmaceutical companies are evaluating significantly more compounds in secondary assays than previously. Indeed, with the continued outsourcing of library chemistry to China, Russia and other countries worldwide, many drug discovery teams are now testing up to several hundred compounds per month in multiple assays. These are not HTS-style assays but rather run the gamut of selectivity, in vitro ADME and toxicology, and PK/PD.

Benn: Theoretically, the use of laboratory automation should have increased productivity within the organisation. So it would logically follow that the impact of laboratory automation would have increased the output from the drug discovery pipeline. However, this has not been the case. The question to ask is, what would the situation be today if laboratory automation had not been integrated into the workplace? This is something that cannot be judged. The anecdotal tales however show that laboratory automation does make a significant improvement in increasing the scope and quality of information gathered across the drug discovery process.

Personally, I believe the true impact of laboratory automation is still to be felt in terms of increased numbers of drugs released onto the marketplace – however this is sure to come.

Keller: Lab automation has developed into a relatively mature discipline in drug discovery through the application of readily available systems either off-the-shelf or customised by external specialist suppliers.

High-throughput techniques have enormously increased the size and quality of the database from which to gain a perspective on factors influencing the decision making in drug discovery.

However, the drug discovery process ought to be based on a good understanding of the underpinning scientific questions and not on the capability of the overall process to generate very large volumes of data based on processed compounds.

Question 5: What do you consider to be groundbreaking techniques occurring in lab automation?

Allenby: I would suggest the capability to integrate and de-integrate equipment rapidly and easily onto and off lab automation would be ground breaking. With the growing cost of liquid handling equipment and reader technology, it is becoming increasingly unlikely that labs will have multiples of each piece of equipment for screening and follow-up work. To fully utilise expensive equipment in an idea world, the equipment should be mobile and “plug and play” in nature around a lab robot that can be moved between labs. The integration process should be a simple, physical “docking” of equipment to the lab automation with instant recognition of the equipment by the software and recall of the loading and unloading positions for the plate. This would allow greater flexibility for expensive lab equipment often under-utilised.

Alton: The most exciting developments in lab automation are developments on microfluidic systems. The key concept is that the sampling platform and the analysis system are integrated on a “chip”. Very complex assays can be performed and reagent consumption is minimised. This leads to enhanced throughput with lower costs.

Benn: There are two areas I have observed showing innovation in recent laboratory automation:
a. Efficiency: analysis of efficiency and workflow has long been an integral part of manufacturing; six sigma, kaizen and lean manufacturing are typical practices in this industry and others. This has begun to be adopted in drug discovery research as applied to processes controlled by laboratory automation.
b. Miniaturisation: today, it is possible to easily and reliably dispense picolitre volumes of liquid; 1536 assays are routine and we can process 1000s of spots on a slide for genomic analysis. The miniaturisation of assays has become so routine and simple now, we tend to forget that even 5 years ago – a quick and simple machine to dispense 500nl was not easily available. I believe this will continue to develop, with the possibility to routinely screen a single cell in a microtiter palte.

Fransson: While Vibrational spectroscopy is not new or groundbreaking, it has increased its area of application. This is due to the new generation of affordable, smarter and smaller instrumentation. It is now a matter of finding meaningful areas of application, such as in-process control and non-destructive measurements of end products, eliminating the need for costly wet chemistry. *Ultra Performance LC™ is potentially a groundbreaking technique, primarily in applications where chromatography has been the rate-limiting step, such as at-line analyses. In addition, formulation development access to expert systems and reliable predictions tools has shortened the development time considerably.

Keller: The last few years have almost exclusively been a period of evolution in lab automation. As a result our focus is more on the ‘technologies’ or ‘applications’ for which the automation is used, for example, Acoustic Dispensing (Labcyte, EDC): non-contact, tip free dispensing of nanolitre volumes.

Question 6: Can you explain how the relationship between vendor companies and their customers helps the development of new technologies in order to meet specific needs?

Allenby: Until recently vendor companies have relied more on their misunderstanding of the drug discovery business than on understanding it. I think project management on both sides is important, with frequent, open and frank discussions on technology development and an honest sharing of problems and issues on both sides. Involvement in product development at a very early stage is vital in providing laboratory equipment and especially software designed with the scientist in mind.

An appreciable understanding of the science is also required by the vendor, either through close collaboration or by alpha- and beta- testing of the equipment by the Pharma. I would suggest that a Pharma science and technology “think tank” should be established, involving creative thinkers from Pharma, vendors and academia to discuss current issues in drug discovery and consider possible solutions through active collaborations sponsored by Pharma.

Alton: Certainly the most fruitful area of collaboration has been in the evolution of the small liquid handling workstation. Biopharmaceutical companies are less interested in spending large amounts of capital on extensive robotic systems. Rather, individual work groups are benefiting from the smaller bench-top instruments that are relatively inexpensive, flexible and are easy to use. The number of vendors offering these systems has grown tremendously in the last few years and vendors are striving to meet their customers’ needs in order to provide key capabilities out of the box without requiring customisation.

Benn: It is my opinion that the relationship between the customer and vendor in the laboratory automation industry sector is crucial. In general, I feel that this relationship works well – indeed if a supplier does not solicit and listen to feedback, I would be unwilling to work with them.

Discussions with vendors on the specific requirements my organisation faces are always welcomed. However, the organisation making the request must also understand that they are but one customer in the portfolio of customers in the vendor’s order book; therefore the vendor needs to balance out the feature requests with the available resources.

Finally, innovations from vendors can also drive their customer’s processes. For example, innovation of applying 2D barcoded sample storage tubes has resulted in the possibility of SMEs in the biotech industry to maintain high-quality traceable sample stores with a much reduced effort.

Fransson: A strong relationship between lab automation vendors and end-users is needed in order to create meaningful technology. In contrast to general consumer electronics, engineers are typically not lab automation users and cannot understand every aspect of its use, particularly in a GMP environment. Users on the other hand, need to be updated on technical achievements, but the actual benefit for the lab can only be fully assessed through sustained dialogue.

Keller: The quality of the interface between scientists and engineers is critical. This interface may lie in the pharma company or between the vendor and pharma company. In the case of the latter, an established relationship is vital to ensure correct interpretation of scientists’ needs, into an engineering science requirement.

Current perception is that the customer, be it GSK, supports vendors to develop a product that either performs below expectation once delivered, or when it is a successful product, finds its way very quickly into other competitor sites.

Question 7: What, in your opinion, are the current limitations within lab automation?

Allenby: The mindset of people! Too often lab automation is used when it should not be, or not used when it should be. How many plates would you screen manually with a pipette before you would consider automating it? What would you use? A Tecan, Hamilton, Beckman? Too often lab automation from fully integrated robotic systems to workstations are used solely within HTS and Hits-to-Lead and nowhere else in the organisation. Why? I think our expectations outweigh the reality of the reliability of lab automation. We expect 100% reliability. When it does not work we reject it for more simplistic approaches. Again we have to ask why? Other industries running complex and highly changing process in an automated manner do not expect this outcome. We need to re-evaluate why we use lab automation. Furthermore, we need to consider the infrastructure required to manage and maintain lab automation internally. In a budget constrained environment, the increasing costs of service and maintenance contracts from vendors can be prohibitive to investment.

Alton: The primary limitation for lab automation is the reduction in capital budgets by biopharmaceutical companies. As companies seek to reduce their bricks and mortar footprint, less money is available for equipment purchases. Thus, automation vendors must respond with instruments that can flexibly adapt to multiple assays and at a much reduced cost of acquisition.

Benn: The main limitation in the industry right now is cost; the equipment we are using remains very expensive to purchase. I understand that this is partly driven by the lower number of units sold per product release (when compared to other industry sectors), therefore causing the money spent on research per unit to be much higher. I believe that smaller, more focused machines to perform a specific task could be developed with a lower cost of research and manufacture – this saving can be passed onto the customer via a reduced price.

In addition, interoperability is an issue. The marketplace for laboratory automation consists of a few big players and a number of smaller more specialised vendors which is a fairly typical situation.

Finally, process control; most organisations cannot quickly and easily see what their machines are doing and when samples are moved through their processes. Compared to other industries such as manufacturing, packing and logistics, this is an area at which we are under-performing and I believe we need to improve.

Fransson: There is limited practical experience within the industry of how risk management can be applied to GMP lab automation projects in order to allocate resources within a project where they are most needed. Moreover, the industry needs to expand its knowledge of working with Quality by Design (QbD) in order to continuously improve, for example development and production methods, without having to change filed registrations with the regulatory bodies.

Keller: The primary limitations within lab automation include; the high cost pressure in the industry, the fact that the SBS footprint remains unchanged and insufficient harmonisation and optimisation of the system to the process.

There is a shortage of experienced resource in project management and systems integrators for the implementation of cGMP/GxP automated solutions.

Most high capital cost automation has in-built down time, during which, sample processing configurations are programmed at a user interface limiting the equipment utilisation. This is a blind spot in design.

In addition, better strategic and economical impact studies are required prior to the start of an automation project.

Question 8: Looking ahead to the next five years, what further advancements do you envisage in this area?

Allenby: I would suggest that apart from improvements in error handling, the field of lab automation is established and any further advances will not bring equivalent returns. Current bottlenecks appear to be the delivery of very low nanolitre volumes of compound to screen in a timely manner to prevent evaporation and degradation. The answer to this may come from physically connecting the compound management robot to the screening robot to generate a more seamless, integrated and continuous delivery of screening plates. Or, alternatively, a more innovative approach to the automated removal of seals from plates during a screening schedule.

The challenge now facing the Pharma industry is understanding how our targets interact in a complex biological system with at best 30,000 components (genes), modelling each interact and determining how we can screen molecules to influence these interactions. I suggest we concentrate our efforts, both vendors and Pharma, into developing technologies that allow us to screen our targets in a more physiological manner.

Alton: The miniaturisation of instrumentation and increased density of assays will drive new concepts in automation platforms. For example, it is now commonplace for cell biology groups to run multiplexed cell assays in 384-well format. Vendors that can supply low-cost instruments that can handle many different assay formats and with an easy-to-use interface will dominate the market.

Benn: One area which will become increasingly important (and therefore requires development) is process control.
Another area which will improve is miniaturisation. I would expect the default format for high-throughput screening to be 1536 – not just in the highly specialised screening areas but in smaller, more focused areas.

Finally, I see advancement in the acceptance of laboratory automation in the scientific workplace. People entering the biotech/pharma industry are now growing up with laboratory automation – it is no longer a novelty. Therefore as these people move forward with their careers, they will be the driving force to really push the boundaries of productivity; utilising laboratory automation as a major tool with which to do so.

Fransson: Increased experience among vendors, end-users and QA personnel when applying risk management and QbD, will increase the number of successful implementations based on this structured way of working. The industry will then be able to establish best practices trusted by the regulatory bodies; practices that will be cost effective and decrease the number of deviations as well as the need for corrective actions.

Keller: There are many improvements on the horizon within lab automation, such as the development of sophisticated on and off line analytical instrumentation and closed-loop controls to enable “self-learning” processes. Also, there is the impact of nanotechnology on screening and analytical sciences, as well as compound management and the application of genomic information to identify and validate new therapeutic targets.

I also expect advancements to include a well defined data management strategy including systematic storage and retrieval methodologies, with a defined archiving and retention policy, along with greater standardisation on equipment, components and consumables to reduce cost.

Acknowledgement

The author wishes to acknowledge the contribution of Dr. Ian Hudson and Dr. Chris Scott of the WW Applied Technology group at GSK, in formulating this.

Key:

TS: High Throughput Screening
FTE: Fully Time Equivalent – a single resource
GPCR: G-Protein Coupled Receptors
FLIPR: Fluorescent Imaging Plate Reader