April 15, 2006 (Vol. 26, No. 8)
Kathy Liszewski
Optimization, Statistical Analysis, through to GMP
Developing and validating appropriate biological assays can spell success or failure for biopharmaceutical products. Whether using live systems or biologically derived reagents, scientists must carefully design, appropriately use, and exactingly analyze data.
At an IBC conference on bioassays in Carlsbad, CA, next month, several experts will discuss key strategies for developing and validating bioassays, including optimization of each step, employing the most accurate statistical analyses, and scrutinizing the long-range ability of the assay to carry through to GMP.
Most scientists are not statisticians and most statisticians are not scientists. Bridging the gap between the two worlds can be challenging. Michael Merges, manager, bioassay services, Cambrex Bio Science Walkersville (www.cambrex.com), offers advice on closing that gap.
“Our group develops, quantitates, and validates bioassays. We also transfer in bioassays from various stages of development by other companies. I needed a tool for analyzing dose-response curves, and I was frustrated at the complexity of most programs. Then I found a white paper on a product called JMP. I tried this program and found it to be extremely user friendly, providing parallel line analyses with the click of a button.“
JMP, which is available from the JMP business division (www.jmp.com) of SAS (www.sas.com), also includes programs for data management, session management, statistical coaching, statistics visualization, and split plot design.
Merges utilizes JMP for fitting dose-response curves. “Bioassay curve analysis is a widely accepted method to assess assay suitability. These types of analyses of cell-based bioassay data consist of plotting serial dilution curves, comparing an unknown compound with a reference standard.
Parallel curves are often nonlinear but similar and result when an unknown compound and reference standard work by the same biological mechanism. So, toxicity, for example, is directly proportional to the concentration of the agent. The asymptotes, or the minimum and maximum toxicity, produce a curve with a sigmoidal pattern. Comparisons of curves determine relative compound potencies.“
JMP also can evaluate curves whose shapes are dissimilar, indicating that the mechanism of action is not the same. “Overall, JMP provides a useful data-interpretation tool,“ Merges reports. “It also adds to the efficiency and simplicity of assay transfer. Our technical staff can use and develop bioassays and, down the line, I can easily transfer the information to QC groups. It is easy as well to transfer to customers who may wish to further develop and validate assays.“
Potency Assays
A challenge that scientists face during development of biological assays is accurate determination of a therapeutic´s potency. “Potency assays provide a quantitative measurement of product quality, manufacturing consistency, product stability, and comparability after a process change,“ according to Xiaochun Qin, scientist, PDL BioPharma (www.pdl.com).
Qin notes that PDL BioPharma develops two types of potency assays. “Binding potency assays can be used to characterize the activity of the product through binding to its specific receptor. For this, we use enzyme linked immunosorbent assays (ELISAs), flow cytometry, or surface plasmon resonance technology (Biacore). We also develop cell-based potency assays (bioassays), which mimic the in vivo biological activity of the product.“
But potency assays are not the entire story, according to Rich Murray, Ph.D., senior vp, CSO, and CTO. “In addition to potency assay development, a variety of other assays are important for the overall successful development of products. For example, in the case of an antibody therapeutic, it is important to test for immunogenicity.
“These assays are designed and implemented to determine if there is an antibody response against the drug, which is itself an antibody. If so, assays are further refined to determine whether the antidrug response inhibits the binding of the drug to its intended target or whether this occurs to a different portion of the drug.
“It is also important to have assays that can accurately measure the levels of therapeutic drug in the circulation of treated subjects. In the case of antibody therapeutics, the challenge is to be able to accurately and specifically detect low levels of the therapeutic antibody in the midst of high amounts of other antibodies in the circulation.“
Dr. Murray suggests that biomarkers also are becoming increasingly important for monitoring and understanding actions of a therapeutic. “Multiplex assays that can monitor a group of interesting biomarkers all at once are being developed and used in clinical trials.“
Modular Optimization
Development of robust and quantitative cell-based assays represents a challenging task for any laboratory. Evaluating and optimizing every aspect of an assay early on in the development process will provide more bang for your buck, according to Bhavin S. Parekh, Ph.D., senior research scientist, Eli Lilly and Company (www.lilly.com).
“When companies set up cell-based bioassays they need to know the key limiting factors,“ advises Dr. Parekh. “Optimization is a critical step prior to initiating validation studies. Two of the key parameters that should be optimized are cell numbers per well and duration of the assay. How you change these can greatly affect the signal/noise ratio and the overall quality of the response. In addition, it is also valuable to optimize plate layout, evaluate consumables and reagents, such as plates from different vendors, and different lots of fetal bovine serum.
“But trying to optimize the assay one variable at a time can be a daunting task. Instead it is better to optimize the majority of the parameters via a DOE, or multivariate design of experiment.“
Even the most optimized assay must allow for variations, however. “Biological systems are always subject to change,“ Dr. Parekh says. “Responses may vary depending upon a number of factors. For example, one specific variable, such as the EC50, which is the concentration of therapeutic that provides half maximal activity, can vary considerably with changes in assay conditions and reagents, thus affecting the potency of the sample.
“Finally, the ultimate goal for validating assays is being able to demonstrate the similarity in curve response of sample and reference standard. So, it is important that the reference standard is representative of the final API. Optimizing these and other criteria are key aspects of assay development, especially if the assays are to be used as quantitative tools.“
To obtain optimum performance of cell-based assays, it is critical to employ process controls at each step, says Gwendolyn M. Wise-Blackman, Ph.D., senior manager of cellular technologies, Cardinal Health (www.cardinalhealth.com).
“We are a contract development organization,“ Dr. Wise-Blackman notes. “Our customers all differ as to the extent to which they´ve optimized their assays. Our job is to take their basic method and see that it will carry through to the GMP environment. To do this we implement process controls and look at long-term issues. So, for example, we assess ruggedness, repeatability, accuracy, linearity or parallelism, range, and specificity of assays.“
Dr. Wise-Blackman points out that all steps from simple to complex are analyzed. “We initially establish cell-culture conditions, such as the best vendor, the percent serum, level of carbon dioxide, and even how frequently the incubator door can be opened. Little changes can have big effects. We had one customer who recommended a cell-based assay, yet when we tried it, we got different results than expected. We found that the critical variables were changing culture conditions carefully to limit density, controlling for passage number and the need to systematically subculture cells two days prior to testing. This example illustrates how subtle culture conditions could critically impact results.“
It takes time to map out and optimize assay conditions. “Although it´s fairly time consuming, optimizing conditions early on is the best way to go,“ she says. “Also, we will work with customers to be sure that the cell line chosen is the best for their studies. One wants to choose a cell line for assaying that best replicates the biological mechanism impacted by their product.“
Overcoming Hurdles
Scientists developing bioassays often encounter a number of biological hurdles. Interacting with others in the field may help solve problems, according to Kelly Oliner, Ph.D., senior scientist Amgen (www.amgen.com).
“It is always beneficial for us to discuss our assay development efforts with scientists outside of Amgen. In this way we can discuss how we overcame difficulties that we have encountered, especially developing assays using adherent cells. Our bioassay group uses multiple technologies, depending on the biology of the therapeutic molecule. For example, we develop cell-proliferation assays, reporter gene assays, apoptosis assays, and ELISA-based receptor-ligand binding assays.“
Dr. Oliner indicates that many parameters must be optimized. “For cell-based assays, we tend to focus our optimization efforts on cell-maintenance procedures, cell density within the assay, plate format, and incubation times. To interpret the data, we employ multiple data-analysis software programs. For linear parallel line analyses we utilize an in-house controlled spreadsheet. For nonlinear dose curves we use third-party data-analysis software.“
Ultimately, most Amgen assays are transferred to a GMP-testing laboratory. Dr. Oliner reports that at this point, critical needs include analyst training and implementation of best practices, such as cell maintenance, counting and handling procedures, and pipetting techniques.
