AUTOMATION OF SAMPLE PRAPARATION IN BIOANALYTICS FOR HIGH-THROUGHPUT, ACCURATE LC-MS/MS ANALYSIS AND LABORATORY INFORMATION MANAGEMENT SYSTEM

Developing high-capacity sample preparation systems and strategies are of key importance in providing breakthrough in the time required to develop a drug by increasing the number/time of analyzed samples. This objective must be achieved without loosing quality within the obtained data and it is of paramount importance in Pharmacokinetic/dynamic studies. The aim of this PhD project in Analytical Chemistry is to automize the manual sample preparation processes for LC/MS analysis using high throughput techniques, and to insert the developed process in the frame of a pharmaceutical bioanalytical process managed by a LIMS system. Sample preparation by protein precipitation is routinely used for removal of matrix component from biological fluids (typically plasma) prior to analysis. It is a generic sample preparation technique, applicable to a broad range of analyte types. Conventional protein precipitation is carried out in vials or collections microplates, with subsequent centrifugation and supernatant removal. Recently, filterplate precipitation has increased in acceptance, because this approach significantly reduces manual liquid handling steps and is readily automated. Automate an assay is important to increase process reproducibility and often the throughput in the laboratory make free from time-consuming repetitive tasks and gave more time for creative thinking. Automated systems will not automatically improve the quality of pipetting and yield. Only in the case of a well-defined assay with a protocol well adapted to the automated process, yield can be expected to be equal or better than in the manual process. Some assays require manual steps that require close eye-hand interaction: e.g. moving the tip of a droplet and aspirate following the droplet. In other words: mimicking complex manual processes can be a challenge even with a robotic workstation’s technology that is using advanced monitoring features to imitate such manual operations. Most liquid handling workstations have a similar hardware architecture. They consist of a deck that is the workbench, where labware with reagents and samples are placed, a pipetting arm that holds the actual pipetting units. The arm moves across the deck so that the pipetting units can reach the labware. There are individual pipetting units as well as pipetting “blocks” in 8-well format, 96-well format or 384-well format. The main and basic parameters that make Hamilton automation so efficient are Liquid Level Detection (LLD), Monitored Air Displacement (MAD), CO-RE, Total Aspirate and Dispense Monitoring (TADM) and Anti Droplet Control (ADC) Technologies. With this preliminary remarks and with a good skill of Microlab Star Vector Software it could be possible to adapt and to perform laboratory methods to increase processes reproducibility and the throughput in the laboratory. The first aim of my project, after having learnt how to use the software and how to adapt the methods to the robot ability, has been to perform a Protein Precipitation Method with a high level of automation. I used a solvent first method carried out by 96-well filtration plates. First I had to adapt the labware to the method, then I had to create the steps of the method with the Software Vector. All the steps, including mixing and vacuum generation could be performed into the robot labware, where the degree of automation for this technique is very high. The method is linked to a System Layout, which at the beginning is an empty environment. I mapped the real environment in the software by adding first a MICROLAB STAR Line instrument to the System Layout. The Instrument is represented by a Deck Layout, that is the graphical illustration of the worktop of my instrument; it contains all the informations about the labware used and X/Y/Z coordinates of positions. I can use on my deck all kind of Labware as tubes, microtiter plate, and reagent through ready made patterns present in the the software. If labware of a different or new kind must be defined, I can use labware editor for any help . The labware describes the geometry of objects, which can be dealt with as a whole, such as the wells of a microtiter plate, or which can be combined on the deck, such as a carrier holding several plates. Then I translated the protocol method in several steps to be performed by my robot. I had to choose the best parameters to reach the best performance. One of the most important parameters is the way to dispense. At the beginning I used a “jet” dispensation but in this way I could not obtain the expected results. So, step by step, changing every time all the parameters that gave me the wrong results, I reached the right configuration on labware and layout, enabling to perform for (the) first (time) correct calibration curves. Each method remaining constant in his performance could be different in solvents and volumes. The concept of Liquid Class is important. A liquid class is a set of parameters determining the most appropriate aspiration and dispense performances of the pipette for any given liquid, tip type and dispense mode. For all aspiration and dispense steps I must select the most appropriate liquid class. I could find several standard liquid classes in the software, but they are not enough to fullfil all the needs present in my methods, since sometimes there were particular liquids or liquid mixtures not covered by any ready-made liquid classes. In the section “Liquid Details” I can set the parameters I need such as the flow rate that correspond to plunger speed for aspirating, dispensing and mixing, the air transport volume, the over-aspirate volume which is a kind of pre-wetting volume, etc. In the section “Correction Curve” I can introduce the corrected values calculated with gravimetric tests. The “corrected volume” is the volume that actually needs to be moved by the plunger for this purpose. In aspiration or dispense steps, the “target volume” which will be actually dispensed into the vessel must be entered. For example a corrected volume of 107.2 µl for a target volume of 100µl does not mean that 107.2µl of liquid will be dispensed. The high flexibility of the liquid classes allows to pipette any liquid with high accuracy. The liquid classes I have calculated for the methods I utilized since now (MeOH-ACN 50-50; MeOH-H2O 50-50; ACN-MeOH 75-20; ACN-Formic Ac. 15-85; etc.) have good characteristics of accuracy and precision in dispensation more than the one requested. I performed several tests to reach the right mixing time and rpm for shaking the plates. I made the same tests about filtration on the Vacuum Manifold. I found the values which made me obtain the same results of manual performances. Sample preparation using automated or semi-automated Protein Precipitation Technique is a cost effective and reliable means to reach throughput goals. Although I always remained at the instrument when methods were run, a workstation yields rapid sample preparation times. An entire plate can be processed in under 10 minutes for the actual liquid handling steps. Additional time is necessary to prepare the sample plate and perform the off-line manual processing tasks. The goal of this application is not to prove that the precipitation plates work, but to show that the plates can be automated. The use of filtration plates offers the advantage of eliminating the manual off-line step of mixing and centrifuging, as used with a collection microplate. In the well the precipitation occurs immediately and protein is trapped on the surface of the filter in each well. Filter Plates Sirocco The Sirocco 96-well filtration plate is an update in the automation and throughput of protein precipitation in drug metabolism/ADME/toxicology labs by providing a simplified means of sample preparation. The plate combines 96-well plate tip design features with proprietary membranes that result in rapid "in-well" sample preparation methods. This plate contains a unique filter system, a sealing cap mat and a patented valve technology designed specifically to allow “in-well” processing which prevents clogged wells, cross-talk or leaking during use. Is possible applying a vented cap mat, using a cap roller to ensure secure, uniform sealing. This will prevent leakage and cross talk during mixing. With Sirocco filter plates there is no need to remove the vented cup mats or valve tips from the plate before filtration because the valves and vented cap mat are designed to open under vacuum to allow controlled flow during filtration. Mix time 0.5-1 min at medium setting, filtration 3 min 10” Hg. Excessive mixing could cause cross talk and under mixing this will cause clogging or a cloudy filtrate. Whatman Fast Flow The plate is made with 2ml, 96-well, rigid glass filled polypropylene which make the plate both robust and chemically resistant. The plate contains specially formulated dual membranes with two distinct layers. The top layer acts as a prefilter to remove coarse particulates. The bottom layer is oleophobic for retaining the well contents without dripping. This provides a final filter for removing fine particulate matter when vacuum is applied. (Reference Application note Whatman PPT). In January, after the optimization of the calibration curve tests, I performed the Hamilton StarLet Validation making use of an internal validation protocol. I made a parallel validation of the method using the two different filer plates described above. The aim of the present study is to perform, at the RBM Preclinical Bioanalysis Lab., the validation of a LC-MS/MS method, for the quantitation of a new chemical active entity in Na-heparinized dog and rat plasma samples. The Full Validation will be performed to determine this new chemical active entity in Na-heparinized dog and rat plasma samples. Detailed description of LC-MS/MS procedure is described in the Test Method PBTM. Spiked plasma samples will be prepared and frozen at –20 ± 5°C before starting each validation program, in sufficient aliquots to complete the study. Spiked samples (Low, Medium and High SSs) will be prepared in the same way as QC samples as described in the test method and used for the following test. Validation, During Full Validation the following parameters will be evaluated: Results Even if the results in Run terms were in behalf of Whatman plates I decided to utilize for the future experiments of pharmacokinetic and toxicokinetics Sirocco filter plates because in term of recovery demonstrated a better performance. Results and discussion Linearity of the calibration curve (over a concentration range of 1-250 ng/ml), as measured by r2, was comparable for both plates. The precision of calibration curve parameters, filtration results, mixing results, inter-day precision etc were comparable for both the PPT filter plate. Only the relative recovery of analyte and Internal Standard using Whatman plates was lower of 16%. In parallel extraction assays After having optimized and validate the system, in June I begun to perform the real samples’ extraction comparing them with manual extraction in real time. Studies In August I performed my first studies (GLP and NO GLP) for Merck Serono S.A. • No GLP Pilot toxicity study monkeys treated by oral route • GLP 4 Week Preliminary Combination In The Monkey Followed By A 6-Week Recovery Study. Second year. After the development and validation of the automated PPT I performed many studies for Merck Serono S.A. • RE8500 TK monkey • RE7930 Phase I reanalysis • RE9030 PK dog • RE5820 TK dog • 0326-2008 TK monkey • RE9840 DRF monkey • RF0150 TK monkey and incurred samples. However PPT technique offers minimal selectivity as it only removes gross levels of proteins from a sample prior to an analysis. In contrast other techniques such as SPE offer significant benefits in terms of selectivity/sample cleanup, but the technique often requires moderate to extensive levels of expertise, multiple steps and time for adequate method development. LLE I developed a fully automated high-throughput liquid-liquid extraction (LLE) methodology for preparation of biological samples using two 96-well recovery plate (1ml and 2ml) according to the methodologies used in the LC/MS/MS Lab instead of the possibility to use a plate filled with inert diatomaceous earth particles which is very expensive and even somehow more toxic. In traditional LLE, a compound partitions between two immiscible liquid phases, usually an aqueous sample and an organic solvent, based on its affinity for each of the liquids. The manual procedure is very difficult and time-consuming. Manipulation of organic volatile solvents is not easy and has safety issues. Moreover good and reproducible mixing of immiscible solvents is not easy to achieve and quantitative transfer of the organic layer for maximizing recovery is really very difficult to do manually. The methods I adapted to the automatic workstation utilize two different extraction solvents: methyl-t-butyl ether or ethyl acetate. Both these solvents did not suffer of the formation of the irregular emulsions between aqueous/organic interfaces and possessed the appropriate polarity to achieve the desired recovery. This approach solved the emulsion formation problem, and permitted the method to be automated using standard 96-well plate technology. Manually each sample is extracted in a 3ml tube, each step is very time-consuming. Time required only for mixing 20”x96=1920”=32’. When TBME is used the bottom of the tube (aqueous) is frozen in N2 and the organic solvent is transferred in 96-well plate 1 ml with risk of cross-contamination. With ethyl acetate the manual transfer is more difficult for the impossibility of bottom freeze, it is completely manual and for this reason not very reproducible. With the robotic workstation the main steps are the extraction and the transfer after extraction. For the first step it is important the mixing settings and to avoid vortex mixing and increase the surface/volume ratio, an aspirate/dispense cycle is repeated 15 times. After this step the organic layer is transferred to the collection plate, to recover the maximum volume the aspiration height has to be 1 mm above the plasma layer surface. Organic solvents used are not polar. Not conductive (PLLd) Pressure level sensing is the only way to determine the level of non-ionic liquids. Contact surface is low and needs to be enhanced by using Vortex mix: this easily causes emulsion formation. Such a problem can be solved in some cases by using 96 wells diatomaceous earth plate, but this is not a fast procedure and requires several steps and costs. Dripping may occur while pipetting organic solvents, with the Anti Droplet Control (ADC) Hamilton can pipette any kind of organic solvents. Time required for processing 96 samples is about 20/30 min while in manual procedure 32 min is time required only for mixing. Using this procedure I performed a part of an important Merck Serono study: CLARITY CLAdRIbine Tablets Treating MS. Cladribine tablets could become the first oral, short-course disease-modifying therapy available for people living with relapsing MS, as all disease-modifying therapies currently approved for the treatment of MS are injectable or infusion therapies. Merck Serono submitted a marketing authorization application to the European Medicines Agency (EMEA) for Cladribine Tablets in July 2009 and will submit further registration applications for Cladribine Tablets in other countries, including the United States, during Q3 2009. The results I obtained using both organic solvents for automated LLE in different studies are very reproducible, not time–consuming and less expensive. The only steps which required attention in settings are mixing and transfer. LIMS Watson is a highly specialized protocol-driven LIMS specifically designed to support DMPK/Bioanalytical studies in drug development. The system was developed with input from major pharmaceutical companies, and its success is a direct result of its ease of use and the high level of service offered to assist in implementation. Watson is installed in 18 of the top 20 global pharmaceutical organizations, and is widely used in leading biotechnology and contract research organizations worldwide. Bioanalysis is an integral part of clinical and pre-clinical drug development. Efficient assay validation, bioanalysis of samples, instrument interfacing, sample tracking, and reporting of results are key steps in progressing a compound through R&D activities and into regulatory submission. Watson includes key functionality specific to the bioanalytical laboratory, including flexible protocol-based study design, assay/method standardization and management, integrated sample tracking and a configurable reassay decision tree. Watson includes more than 70 built-in interfaces to LC/MS, HPLC, ELISA, RIA, ICP/MS, multiplex and other instruments. It also supports a wide range of pharmacokinetic/toxicokinetic calculations. The acknowledged standard for bioanalytical LIMS, Watson facilitates efficient study design and data transfer between researchers, enabling them to benefit from improved operational efficiency. Using LIMS fuctionalities and reports, I developed an Excel file able to manipulate sample data from LIMS into the Hamilton data format: sample dilutions, position and sample identification data, and sample sequences are now flowing from the LIMS to the Hamilton workstation enabling sample automatic and error-free sample handling and processing. Dried Blood Spots. In the future I would like to adapt to the robotic workstation the use of a new technology of sample shipping and analysis. It is the DBS, a novel approach that has been developed for the quantitative determination of circulating drug concentrations in clinical studies using dried blood spots (DBS) on paper, rather than conventional plasma samples. This technique represents a very easy way of collecting, shipping and storing blood samples. Blood from animal or human is directly spotted onto collection cards that are air dried and stored desiccated at room temperature. For the DBS analyses, a 3mm diameter disk is punched from the centre of the DBS into a clean tube. DBS simplify process, there is no need to centrifuge, sub-aliquot, freeze and defrost samples. It also reduces samples volume from 300µl to 10-20µl consequently reduces the number of animals to be used. SPE. The last sample preparation technique I would like to adapt to the Hamilton workstation is Solid-phase extraction (SPE). Solid-phase extraction is a versatile and selective method of sample preparation in which analytes are bound unto a solid support, interferences are washed off and analytes are selectively eluted for further workup and analysis. My goal in developing and using 96-well SPE methods is to achieve high throughput with time and labour saving. Efficient use of automation is the key to achieving this goal. Third year Solid Phase Extraction (SPE) High-throughput solid phase extraction utilizes 96 well microplate formats. Although the samples can be processes manually liquid handling workstation is preferred; the use of the automatic workstation results in very high throughput. The aim of this part of the project was the automation of the sample preparation in the study for the evaluation of the stability assessment of a mixture of four natural peptides in human blood (derived from Myelin Basic Protein) using the UPLC-MS/MS technique. The study was conducted according to MSR-QMS. This mixture is designed to induce immunological tolerance of the body’s T-cells to key auto-antigens involved in the pathogenesis of Multiple Sclerosis. Oligodendrocytes form a protective sheath, known as myelin that insulates the fibrous cables, or axons, radiating from nerve cells. In multiple sclerosis, the immune system's T cells and B cells attack oligodendrocytes, ultimately damaging the myelin sheath to the point that the electrical signals transmitted by the axons beneath it are disrupted. The method set-up was thoroughly investigated in order to obtain the best SPE conditions. During this phase, different SPE plates and different solvent mixtures were tested to obtain the best results in term of recovery, robustness and reproducibility of the extraction method. On the whole, analysis sampling and sample preparation are the most important steps of analytical procedures. Due to its flexibility, SPE is a multivariate process whose optimization should be approached from a multivariate perspective. For the optimization of this SPE procedure we investigated the selection of the most appropriate sorbent, the design of the SPE bed, the determination of the volume of the sample to load and the determination of the nature and volumes of solvents first used to wash the column and then to elute the analyte. Microelution SPE was used in order to avoid solvent and sample spending. In designing the automation strategy for Solid Phase Extraction there are many parameters that the operator has to consider. Initially has to be found the most convenient SPE condition, than it is important to planning of the most appropriate deck layout which has to be different from the other described before because of the needs of this technique. At this point the microplate can be placed onto the workstation deck for all subsequent pipetting tasks. Once the sample plate has been prepared and place onto the deck the first pipetting action by the instrument is to aspirate the solvents for conditioning the plate. These solvents are sequentially aspirated from a reagent reservoir and dispensed into all the wells of the plate. If needed a vacuum manifold step is performed between. It is important to evaluate the correct measure of the vacuum to apply. It has to be not too strong but not too low. Than a volume of sample matrix is next aspirated from the sample plate, a vacuum manifold step follows. A clean collection plate is placed inside the manifold. In first was used an Oasis HLB µElution plate. It could have been correctly placed in a good position for dispensation and vacuum without moving any plate. In fact this plate has a low high and the axes of the workstation were inside the settled parameters. During the clean up was noticed that there were some percolation problems. For this reason was decided to use Strata X plate but the deck Layout positions had to be changed for the plate dimension. It was used a different carrier to avoid the collision between the tips and the plate. Although the results obtained during stability assessment showed a rapid enzymatic degradation of the four peptides in human blood the use of liquid handling to automate the extraction was successful. The recoveries obtained in the manual and automated extraction procedures and chromatograms are comparable. It shows that all liquid handling steps can be performed by the workstation proving to be quick and labor saving. Dried Blood Spot (DBS) In this technique blood is directly spotted onto collection cards it is air dried and stored with desiccated at room temperature. The process is simplified and the possibility to obtain better quality is higher. It is possible to avoid: • Centrifugation • Sub-aliquoting • Freezing • Defrosting. Samples volume passed from 300µl to 10-20µl, in consequence: • Reduction of the number of animals used. • Reduced costs • Shipping • Storage A new small chemical entity was used to evaluate the possible application of the technique. An NCE (new chemical entity) is a molecule developed by the innovator company in the early drug discovery stage, which after undergoing clinical trials could translate into a drug that could be a cure for some disease. It was evaluated the possibility for DBS technique to bring advantages in Merck Serono PK processes. It was evaluated: • the sample processing procedure in order to find the most suitable type card, the appropriate solvent and mixing extraction time. • the accuracy and precision, using QC samples at three different concentrations, the matrix effect and the recovery. • the reduction of the punched area diameter maintaining the same MS peak area ratio with the purpose to enable the automation of the extraction process. Finally it was investigated if the molecule of interest is equally distributed onto the blood spot. A simple and rapid assay has been developed using whole blood dried onto filter paper cards. The collection of samples is easier, and sample preparation prior to analysis is less time-consuming and requires fewer resources, as it involves the simple extraction of compound from a disc punched from a dried blood spot into LC mobile phase solvent. The method is both precise and accurate, selective and linear over the investigated concentration range with acceptable inter and intra-assay variability. We selected the correct card type for our molecule, the best extraction solvent and the appropriate mixing time. We did not detect the influence of hematocrit (patient A and B) composition. The blood spots proved to be stable during a period of 1 month at ambient temperature. The intra- and inter-run accuracy and precision are within the method acceptance criteria. It has been found that sampling location and punching position on the spot do not significantly affect the precision and accuracy of the assay. We have been able to use 3mm punched spots with 50µl of extraction solvent in order to enable the automation of the extraction process in future. In conclusion we developed a method suitable for analyzing our molecule and DBS technique can be considered as alternative technique for PK/TK Regulated Bioanalytical studies giving results comparable to the standard methods, but with significant savings in time and resources. In conclusion it is possible to observe that through use the Hamilton MicroLab Star liquid handling workstation, different sample clean up techniques has been successfully automated and implemented for several biological matrix samples. This in turn has greatly reduced the intensive labor as well as the possibilities of systematic error associated with the manual volumetric transfers. Significant advantages in terms of efficiency and throughput have now been realized by this automated process while maintaining the integrity of the precision and accuracy.Over 12 studies and 6000 samples have been prepared in the Automation Laboratory since implementation in the first six months. The original goal to automate the sample workload using this approach was met.