Banking on blood

Designing an automated blood fractionation system for the UK Biobank

The demand for fractionated blood has increased year on year, since the mapping of the Human Genome. As a result, the trend towards biobanking as a resource to help deepen understanding in a variety of diseases is growing, as large studies require samples to be processed rapidly and in very high volumes.

Tim Peakman, executive director at UK Biobank, commented: “The required throughput and quality and highly ramified data structure means that this process can only be done in a fully integrated and automated way. We have worked with RTS using modern manufacturing design principles to produce a robust automation platform that will meet the needs of the project in this critical process element. Not only is the data trail absolutely secure and resilient, but there isn’t a single element of the system that relies on brand new, untested technology. Our HTBF Platform is new only in its application and the way in which its constituent mature technologies have been brought together and configured.”

The deal
RTS Life Science won the contract to design and install three automated robotic systems incorporating machine vision for the new UK Biobank. The project follows the successful development of a prototype vision system (patent applied for) by the Company. The first HTBF Platform, will be installed and operational in the third quarter of 2006, by which time it will be fully integrated with the UK Biobank’s software and database. The second and third platforms, required to meet the extremely high throughput demands, will be operational by the end of 2006.

Hosted by Manchester University, the UK Biobank aims to build a major resource to support a diverse range of research that will improve the prevention, diagnosis and treatment of illness and promote health throughout society. The project will follow the health of 500,000 volunteers for up to 30 years, collecting information on environmental and lifestyle factors and linking these to medical records and biological samples. The samples will be stored so that they can be used for biochemical and genetic analysis in the future. UK Biobank is funded by the Department of Health, the Medical Research Council, the Scottish Executive, The Wellcome Trust and the Northwest Regional Development Agency (NWDA).

Paul Downey, director of laboratory operations at the UK Biobank, explained: “No one has ever done anything like this before. We worked out that our programme will require approximately 3,500 tubes of blood to be processed each day. When centrifuged at low speeds, blood separates into three layers, or fractions: the plasma, the buffy coat containing the white blood cells and the red blood cells. As blood degrades over just a few hours, processing needs to be speedy. In addition, the size of the layers varies from person to person and even alters from day to day. The only alternative to an automated system would be a manual one. Even for highly skilled technicians, aliquotting the right amount while ensuring that the other layers remain undisturbed, remains a tricky procedure. Quite apart from the small army of technicians such an output would necessitate, there are also the serious health and safety aspects exposing people to unscreened human blood every day of their working lives. It soon became apparent that there was no machine that would do all that we required, so after consultation and a tendering process, we chose RTS Life Science, because it was the only company that had the combined experience in automation driven by vision systems and life science automation.”

The RTS solution
As this sphere of science is changing so rapidly, the RTS team realised that its system needed to be supremely flexible, so that it can easily accommodate new parameters set by the user. Some researchers require all three layers, but many just focus on a single element. Unlike many other biobanks, the UK Biobank will collect all the parts of the blood, as each part can reveal something different about a human’s health and more detailed assays are likely to be developed in the future. The focus of both teams’ effort was the narrow buffy coat, as this often annular layer is so difficult to collect successfully. A large degree of variation in the volume (and therefore weight) of collected samples is expected. In order to address this prior to centrifugation, the HTBF Platform weighs the samples and then carries out volume sample variation analysis to ensure the centrifuge is correctly balanced and optimally loaded. This is a complex operation whereby the small Staubli TX40 robot loads two racks full of samples on to two separate balances. If required, individual vacutainers are moved between the two racks until the mass of each rack is within 10g of each other, achieving balance with the centrifuge. The system’s main robot, a Staubli TX90, is responsible for ensuring the correct positioning of all the requisite labware and racks, with the smaller TX40 robot loading the centrifuge once the correct sequence has been determined. The centrifuge is capable of holding four racks of 24 tubes.

Some 3,500 tubes of blood will need
to be processed each day if UK Biobank
is to be a success
Experienced in the sphere of machine vision, RTS developed a system especially for this project. A digital camera linked to an ordinary PC, with a special “frame grabber” graphics card, takes two images, while lit by Light Emitting Diodes (LED). This form of light provides a very even field of illumination. In addition, LEDs generate very little heat, thereby helping to maintain sample integrity. This light also eliminates any surface reflection from the tubes. As soon as the tube is taken into the vision system, the bar code label is read and then the tube is immediately rotated, so that the label does not interfere with the reading of the different centrifuged layers. Images are taken two at a time, with the grippers selecting the two chosen samples and pulling them upwards to a light proof box. Once the images are taken, they are gently lowered into the rack before the next pair is selected.
                                                                
Unique software first analyses each individual image and then combines and refines the different data supplied by each image to build up an accurate representation. It is then able to determine the exact boundaries between the layers of centrifuged blood. The layers are then converted into liquid handling protocols and mapped to destination tubes. With an eye to the future, the system can handle 96 or 384 way tubes.

Before the robot aliquots the layers into arrays of 96 1ml tubes prior to storage or preparation for assay, the caps on the vacutainers must be removed. The caps are required throughout the process to avoid contamination. Any agitation would render the results from the vision system void. The caps are therefore carefully removed by an RTS designed unit.

Many of the biomarkers within biological samples degrade very quickly at room temperature. To minimise this sample degradation, the HTBF Platform is housed within a cold room maintained at a steady +5oC. Many of the current robotic liquid handling devices do not operate at +5oC, so these devices are positioned on the outside of the cold room. Following studies using sophisticated computational fluid dynamics analysis, an air replenishment system draws cold air out of the cold room and across the deck of the liquid handlers, keeping the samples cool during the aliquoting process. By keeping air flowing round the system, there is also no chance blood aerosols can build up. If the samples are not required immediately, they are sent for storage at either –80oC or in vapour phase liquid nitrogen.

Traceability and quality control are assured, as unlike people, robots do not get bored! The specially developed software is based on RTS’s SPRINT scheduling software. It not only controls the vision system, but interfaces to the users’ LIMS software and links into a database for recording and identifying all the sample information.

Adrian McQuillan, RTS’s project manager, commented: “The speed of aspiration is critical. Plasma is the fastest of the three, because it is less viscous, but we cannot risk disturbing the buffy coat, so our system parameters stop just short of it, ensuring that there is no layer mixing and no buffy coat is lost. Shear forces must also be considered for cellular fractions.”

Projected results
David Harding, business development manager at RTS, said: “While one of our HTBF Platforms can process between 100-300 vacutainers per hour, depending on the protocols used, a laboratory technician would require a day to process just 20 to the same quality and with the same level of data integrity. As experts in both robotic integration and machine vision, we are well placed to be at the forefront of this emerging market, which includes not only other biobanks, but also, High Throughput ADMETox, and other biomarker and drug discovery programmes.”

The initial results of the novel vision system show that each of the three layers of fractionated blood can be reliably and accurately detected.

The widespread use of biomarkers found in blood to indicate disease states, is, as is the case in much of this field, still in its infancy. However, understanding of elevated levels of biomarkers for patients with conditions like cancer is surely set to rocket. Without automation to underpin this new research its benefits would be painfully slow to reach the population as a whole.

By Dr. Sean Sales, applications consultant, RTS Life Science