Maximizing compound value to ensure successful downstream discovery operations was the theme of the second “Compound Management, Integrity, and QC Summit,” organized by the International Quality and Productivity Center, held recently in Philadelphia.
In summing up the presentations from the conference’s first day, Jas Sanghera, Ph.D., conference chairperson and commercial director at TTP LabTech (www.ttplabtech.com), described the tightrope compound managers walk, having to balance the needs and interests of the chemists and biologists in an organization. Smaller companies must weigh the benefits of purchasing robotic systems to augment productivity through increased automation while working within the confines of budgetary constraints on capital investments and selecting equipment that fits the company’s needs.
“Don’t go out and automate everything,” cautioned Dr. Sanghera, encouraging companies to purchase modular systems, adding components as needed, and ensuring that the systems can function in an integrated manner and link seamlessly with existing instruments and processes.
According to Elizabeth Grieco, research investigator in the chemistry department at Infinity Pharmaceuticals (www.ipi.com), the basis of cost-efficient production of high-quality libraries is threefold: communication (sharing of data); integration; and use of web-based tools for compound tracking and data sharing.
Infinity creates and sells compound libraries to fund its internal drug discovery program focused on cancer therapeutics. Four groups work together to form the company’s integrated library management team: chemistry (library design and synthesis); analytical (liquid chromatography/ mass spectrometry of synthesis compounds and of full library post production); compound management (CM; process, store, and deliver libraries); and information technology (IT; plan library production, design tracking software, develop CM and QC annotation tools, and analyze library diversity).
All decisions should be made “as a team, driven by data and tools,” said Grieco. “The goal is to take the subjectivity out of library production decisions.”
Complete QC of one sample library at Infinity showed that only about 45% of the library had a target purity of 75% or higher (compared to the predicted and desired 90%). A review of QC testing revealed that the process was marred by inconsistent judgments, subjective decisions, limited standardization, and lack of data and consistency regarding the use of analytical systems and methods.
Changes instituted at Infinity gave the CM group responsibility for processing all pathway tests and QC data and for controlling the flow of data back to the chemists. Additionally, analytical methods were standardized, and recorded and analytical data was preserved in databases.
Grieco identified several specific strategies designed to improve workflow. For example, to minimize lost time for methods optimization, the company uses a small subset of early compounds to develop and optimize analytical methods. For compound encoding, Infinity uses a combination of radio-frequency identification and AC tags—lantern and transtem assemblies—and scans each tag as individual compounds are processed and moved from one microplate to another. After the chemistry group performs synthesis reactions, it sends the QC compounds to the analytical group for LCMS and the QC lanterns to the CM group. The chemistry group then reviews the LCMS results to determine whether the reactions proceeded as planned and to check the yields.
To preserve QC compound tracking during the cleavage steps, the cleavage reaction solutions are transferred from 96-well array plates to weighed, barcoded glass tubes, and then they are transferred to 384-well plates arrayed in four quadrants. The process of arraying, cleaving, and formatting about 10,000 compounds takes approximately three weeks.
High-throughput compound analysis at Infinity involves annotation of all chromatograms using tools and interfaces designed by the IT group. CM cherry-picks compounds based on the annotations and formats them for delivery to the customer.
A subsequent panel discussion, which focused on evaluating the value added by analytical analysis, emphasized the importance, at the outset, of defining the goals of analytical studies—to determine purity, concentration, and/or solubility, for example—and ensuring that the results contribute to the decision-making process.
The panelists were asked how to manage archives of older compounds, how to monitor their quality, and whether to periodically QC entire libraries. The response differed depending on the intended use of the compounds. For libraries used in high-throughput screening it might be sufficient to QC a subset of the collection, perhaps selecting compound sets based on chemical diversity.
“For lead optimization, you might want to know that what you have is the right stuff,” responded Rodney Bednar, Ph.D., senior investigator for pain research and chief drug discovery engineer for the facility for automation and screening technology at Merck Research Laboratories(www.merck.com).
QC analysis is a good idea “before sending a compound forward into development,” said Greg Nagy, Ph.D., research operations manager for materials management at Amgen (www.amgen.com). It is a matter of weighing the cost of a full analytical analysis with the cost and risk of moving ahead with a compound that may not be what you think it is. Consensus estimates for QC testing ranged from $2–3 per sample for in-house, high-throughput LCMS to $5–10 per sample for outsourcing to a CRO.
Dr. Nagy described Amgen’s tiered approach to the organization of libraries based on the purity of compounds. The company is in the process of converting from an archival approach to compound collection to a split storage strategy. One question is when to introduce online, high-throughput analytical technology, either at the point of use or earlier, when building a compound inventory. Dr. Nagy emphasized the importance of having in place a parallel IT initiative to look at the analytical data and provide real-time information on changes in a compound’s structure or activity.
In response to a query about whether to perform IC50 tests on HTS hits in liquid or powder form, Marybeth Burton, associate director for chemical technologies at Schering-Plough Research Institute (www.schering-plough.com), suggested QC analysis of hits in liquid form by the HTS group, and eventually, if the compound moves forward in development, the therapy group would QC the solid form.
When testing an historical library, there is no one ideal LCMS method or set of conditions for all compounds. Developing an LCMS strategy for high-throughput QC analysis “is a delicate balance between throughput and methodology,” said Burton.
Demand for More QC
“Customers are asking for more and more analysis and QC,” said Grieco.
“Because of this increasing demand for compound analysis, we need to invest in higher throughput LCMS systems,” added Burton.
Dalin Nie, head of compound management at AstraZeneca(www.astrazeneca.com), described compound collections as the crown jewels of a pharma company. The goal of the CM group is to deliver compounds with good chemical and physical properties, together with accurate information, all in a timely manner.
Chemical quality depends on structural characteristics, biological relevance, and diversity, whereas physical quality refers to a compound’s identity, purity, and concentration. Many factors can affect a compound’s physical quality, including tracking errors in which IDs are lost or switched, compound degradation, absorption of water, evaporation due to freeze/thaw cycling, or contamination. An initial check of physical quality relies on LCMS for mass confirmation, with a purity limit of >85%.
QC monitoring of a company’s global collection requires the definition and implementation of a system for sample analysis, data sharing, sample selection, handling of failed samples, methods standardization, and crosschecking of failed samples. Nie recommended that all sample processing be carried out in a controlled environment, using fully automated systems and rigorous validation procedures.
At Eli Lilly (www.lilly.com), the chemistry group’s aim is to produce highly diverse drug-like compounds with >90% purity.
“There is no single property to determine if a compound is drug-like. The best approach is to strive for a good distribution of drug-like and non-drug-like properties,” said Rick Loncharich, Ph.D., head of global sample management at Lilly.
Lilly changed its screening strategy to emphasize quality over quantity. “Smart, iterative screening allows medicinal chemists to tailor the diversity and maximize learning,” said Dr. Loncharich. Instead of screening >500,000 compounds in 100 HTS runs per year, Lilly is screening fewer samples iteratively—<10,000 compounds in 1,000 runs/year of iterative panel profiling.
The quality of lead compounds affects the productivity of medicinal chemistry, observed Dr. Loncharich. It requires far more resources to optimize lead compounds with poor property profiles than to start with compounds with favorable physiochemical, ADME, selectivity, and activity properties.
Accumulated structure-activity relationship (SAR) data at Lilly has demonstrated that hits can be transformed into leads with higher potency through only slight increases in molecular weight. The lead-generation strategy should focus on developing compounds with low mean molecular weight (<400) and optimal properties.
When Lilly performed QC analysis of a sample collection stored in plates in walk-in freezers and found that the purity ranged from 70–91% and the average concentration was 60–75%, the company asked, what could we do differently? One change involved storing samples in tubes rather than in plates. Another strategy was to generate a master solubilization stock from which a rack store and a tube store could be made, with the tube store becoming the source of compounds for screening and related studies.
Dr. Loncharich described the use of the Protedyne (www.protedyne.com) BioCube™ laboratory automation system, with a throughput of 3,000 samples/day, for performing sample solubilization in a nitrogen environment. Other automation platforms used for sample management at Lilly include RTS Life Science’s (www.rtslifescience.com)Compact-Store™ tube store, with a throughput of 400 samples/hour, or 6,400 pick-place/day, and storage capacity for 2 million 1.4-mL tubes, and Velocity11’s (www.velocity11.com) BioCel™ , with a throughput of 7,500–15,000 samples plated/day. The TekCel (www.tekcel.com) TubeStore™-225 can accommodate up to 225,000 0.3-mL tubes.
Implementing Lean Six Sigma
In addition to outlining the benefits that automation provides, Dr. Loncharich discussed the implementation of Lean Six Sigma process-optimization strategies at Lilly and the estimated $250 million saved in the first year these methodologies were applied. Lean Six Sigma yielded several key improvements: reduced variability of work processes and a 44% overall improvement in process cycle times, expanded capacity, and reduction of operating expenses.
The main goals of Lean Six Sigma are to identify the causes of and minimize variation in processes and workflows and to limit waste and operator error. Dr. Loncharich also described the Value Stream Mapping tool for identifying which processes to change to reap improvement in the overall workflow. This can be used to guide decision-making on when and how to introduce new technology and instrumentation to reduce downtime, combine steps, and minimize process time, for example.
Husam Fayez, principle research scientist at Wyeth Research (www.wyeth.com), described the advantages of storing samples in DMSO at –20ºC in the dark under inert gas (nitrogen).
Debate over the value of using an argon-purge step to create a plug-over sample has focused on the decapping and recapping steps. Wyeth has adopted an argon-purge strategy to prevent air and moisture from interacting with the samples when storing aliquoted plates in a CyBio (www.cybio-ag.com) CyBi™-PlateSafe unit. Fayez also described the company’s use of PerkinElmer’s (www.perkinelmer.com) MiniTrak™ automated liquid-handling system for filling tubes, REMP’s (www.remp.com) sample store systems, and The Automation Partnership’s Solar™ aliquoted sample-management system, HomeBase™ sample storage and retrieval system, Concerto™ software, and heat-sealed PicoTubes.
In the future, Fayez predicts continued use of –20ºC for sample storage and a trend toward miniaturization with increasing use of picoliter quantities. Storage in a nitrogen environment has replaced air in recent years, but Fayez expects a return to storage under air in the future. He described a change in tube-sealing techniques from a previous reliance on adhesive plastic to the use of rubber caps at present, to heated aluminum pierceable caps in the future. A key shift in compound-management strategy in the future will emphasize sample-location tracking in a centralized database.
Looking ahead, Fayez anticipates throughputs of 50,000 samples/day. QC analysis will include measuring moisture levels in DMSO samples and will focus on monitoring purity and integrity of a select number of compounds using LCMS.
“The goal is improved reliability and unattended operation while maintaining quality and solubility of DMSO samples,” he said. Wyeth conducted a study in which DMSO samples, containing no compound, underwent ten cycles of storage, thawing, decapping under argon, liquid handling, recapping under argon, and freezing. Determination of moisture uptake revealed that water levels increased minimally from 0.1–0.2%
Amgen has made many recent changes in its approach to compound management. The company has substantially modified its IT infrastructure and sought strategies for changing researchers’ mindsets to accommodate change. Amgen completed installation of a REMP storage system this year and expanded its Haystack™ archival storage system. In July it rolled out its new IT infrastructure.
Amgen’s new multitiered storage strategy is intended to streamline operations and enable seamless integration of its global compound collection. It designed its new CM facility to allow for change, for expandable storage/freezer space, to accommodate different types of robotic equipment, and for efficient DMSO liquid handling and waste collection. “The process flow should enable flexibility and redundancy,” said Dr. Nagy.
Lessons learned at Amgen include the importance of gathering metrics and having access to historical metrics. “IT should be enabling, not disabling,” he added. “It should facilitate process alignment across sites and functions and seamless integration of new compound stores and equipment.”