Back to Molecular Med Monthly Articles main page
Multiplex
Assays: An
Interview with Brian Edmonds The advent of DNA microarrays in the context of the Human Genome Project introduced high-level multiplexing to the research market and triggered a shif in the prevailing research paradigm from a hypothesis-driven to a more discovery-based orientation. Subsequent extension of high-level multiplexing to proteins stimulated intense focus on the discovery of new biomarkers, many of which contained multiple analytes and were subject to multiplex analysis. In Multiplex Assays: Technologies, Applications, and Markets, author Ken Rubenstein, PhD, details the landscape of this topic today. Below is an interview excerpt from the report. For more information on this report, please visit: http://www.advancesreports.com/all_reports/2006_63_Multiplex/overview.html Cambridge Healthtech Associates (CHA): Please start by providing a brief background of your position and its relationship to multiplex assays. Brian Edmonds (BE): My position in the company with reference to multiplex assays has to do with my group’s development of assays for, basically, all of Lilly. We are the assay development and optimization group for the various therapeutic areas. Multiplexing became appealing to us, especially from a cost standpoint. In other words, if it costs X millions of dollars to run six single-plex ELISA assays in a clinical trial, and it only costs a few hundred thousand dollars to run one multiplex ELISA for six of the analytes simultaneously, then it’s sort of a nobrainer that you do the latter. So I think that was actually the driving force. Obviously, the data quality has to be the same as for the singleplex assays; you don’t want to sacrifice quality. Our goal was to improve workflow and affect the cost of the assays. CHA: Is it much more difficult to develop a multiplex immunoassay than the corresponding individual ones? BE: The main issue with developing multiplex immunoassays is crossreactivity of the reagents, so one has to first find good antibodies. Sandwich ELISAs are the typical fundamental platform here. So one has to find matched pairs that work very well in single-plex mode, that give you the sensitivity and specificity you need to make a meaningful measurement. Then you start combining these single-plex assays into the multiplex one, and you have to constantly monitor how the performance of each assay is affected when you bring in the next measurement pair. Ultimately, what many people have found is that, at least with the current type of immune reagents used in these types of assays (polyclonal or monoclonal antibodies), you can only get up to about 30 pairs of antibodies before you start to see erosion of the performance of the overall assay. Virtually every group I’ve talked to that does this kind of thing has come up with the same conclusion. So it’s not that it’s any one particular vendor’s antibodies; this crossreactivity just seems to be the nature of the biology of antibodies. CHA: Once you have developed a multiplex assay for preclinical or clinical work, would it be much of a stretch to adapt the assay for diagnostic applications, just doing the validation necessary to get it on the market? BE: That is a complex question. At a simplistic level, yes, but much more is involved in getting a preclinical or homebrew test commercialized. Building the test and using it are different than building the test and commercializing it. CHA: Would those be mainly production-related issues, or would they have more to do with optimizing the assay itself? In other words, would it have to be optimized further for diagnostics? BE: Depending on the envisioned use in the commercial market, perhaps. One has to approach it case by case, but technically, you should be able to translate early-stage assays to a full-blown diagnostic. The earlier in drug development that we can determine if a commercial diagnostic test will be important for a given Lilly therapeutic (now popularly referred to by the buzzword theranostic), the earlier we can incorporate an accepted commercial technology platform into the design of the assay. CHA: Your group and others seem to favor encoded bead arrays over fixed, two-dimensional varieties. Why is that, and what are the advantages of the beads over two-dimensional arrays? BE: If we are outsourcing, in other words, if we are only providing samples and some other entity is actually running the assays, we don’t care what the platform is. We are only concerned about the quality of the data that come back. If we are running the assay in-house, for us the workflow is much more streamlined for bead-based arrays than the planar arrays. We have automation platforms for liquid handling that work with the bead-based arrays, and only a couple of microarray companies have come out with slide washers. I don’t know of any company that has the slide washer hooked up to the actual scanner so that it really is a push-a-button-and-go-away kind of thing. Typically in the pharma environment, we don’t like to do things manually, especially when we have many samples to process. Thus, automation is a key feature. The slide-based arrays don’t really have that, and we would have to build it ourselves, which is something we’d prefer not to do. Both platforms now have similar analyte content, so that’s not really a driving factor; rather, workflow becomes the driving factor. If needed content were only available in two-dimensional arrays, we’d bite the bullet and do that. In the past, we worked with Molecular Staging in sepsis with protein microarrays, and great data were generated. However, we didn’t have to run it; Molecular Staging had built a whole infrastructure to do that in a fairly high-throughput manner. CHA: Concerning the paper you published [Heuer et al. 2005], you have provided a view into multiplex assays in pharma that no one else has presented. BE: It is interesting that we had some difficulty getting the paper through editorial review. They wanted us to take out a lot of things, especially with respect to experimental design and statistical analysis. We said, “No, we think we need to make these comments.” We have gotten a lot of positive feedback like yours though. CHA: Concerning fluorescence background on your two-dimensional arrays of cytokine antibodies, the signal differed between two laboratories, and I was wondering if the high one was so high that it might saturate the detector. BE: The signals coming from the actual spots were much brighter than the background signal surrounding them, so there was no problem with the background causing saturation. The issue was more that the background was not uniform across the slide. So it really speaks to the various algorithms that these slide scanner companies use for background correction, and some of the assumptions they make are flawed. One needs to be smart while doing these kinds of microarray experiments and be sure to do the data analysis correctly. That is not endemic just to protein microarrays; it originally started with the DNA arrays. It is just a problem of microarrays in general, regardless of the analyte. CHA: Is there a background with the Luminex bead arrays? BE: There is background there, but because of the nature of the data acquisition, it is just dealt with differently. With protein microarrays, the problem is image analysis. You need to use spatial information to be able to correct for artifacts. In a virtual bead-based array, it is obviously not two-dimensional. There still is background, but it’s dealt with differently, for example, by including controls with unconjugated beads, labeled beads without biological sample, and so forth. CHA: Quantum Dot was recently acquired by Invitrogen. Does that acquisition make its system more attractive as a possible alternative to Luminex? BE: It could be an alternative if it chooses to implement the system as such, but I have not seen it do that yet. One thing you can be sure of is that technology always becomes obsolete; something inevitably replaces what went before. Will Quantum Dot, in the hands of Invitrogen, cut it? When it has a commercial system available, we will test it. We are very flexible, but we have no idea what the quality of their data is, what the cost per data point is, or what is entailed in making the measurement in terms of workflow, time, and that sort of thing. Also, their system targets moderate-scale gene expression analysis. We do not do that a great deal; I don’t know why. We either do the “Full Monty” with an Affymetrix chip or we do a small handful with TaqMan or bDNA [branched DNA]. However, we do not go into that middle area yet. If a future need arises, we will assess what technologies are available to do that. CHA: In using the Luminex system for clinical studies, where are those instruments located? Do you operate via CROs [contract research organizations]? BE: It is a mixture: Some CROs have them, and some buy whatever we ask them to. If we have developed an important multiplex assay at Lilly and we have determined that it has to be run by a CRO, the CRO partner will acquire a system, get trained, and run it. CHA: Speaking of high-throughput multiplex assays in pharma, do you see this as something that is on an upward growth trend? BE: I think there is continual uptake of the pre-existing content; for example, every year more and more people start doing cytokine assays. So the growth we are seeing is entry into the market. Is there growth among existing users, however? I think we might be approaching a plateau. I spoke at a Luminex workshop earlier this year. As I was walking into the conference room, I expected to see 20 or 30 people there based on what I had heard about the turnout at previous meetings. However, approximately 150 people were there and lots of vendors as well. The organizers were also surprised at the turnout. Clearly, growth is occurring, but when I listened to what people were doing, it was the same old thing. So I don’t see much innovation on new content or applications creeping into it, just more use of what is already there, with more and more people doing it. Companies are not going to innovate unless they can get a good return on their investment. For example, Bio-Rad was one of the first companies into the Luminex arena. It was pretty innovative in the beginning, then not much was happening, and then it started up again. Bio-Rad’s explanation is that its sales are now justifying further investment. So it could have kept on going and cranking out innovative products, but it was not clear that it could sell much. Now it has the sales to support renewed investment in growth of its product portfolio. CHA: A company such as Rules-Based Medicine [RBM] has a fairly limited menu of potential biomarkers using the Luminex technology, which clients can access for biomarker discovery. Is that menu large enough to discover new biomarkers, or do you need a larger selection of analytes? BE: The bigger the fishing net, the greater the probability of catching something interesting, but what RBM has is better than nothing. One big advantage from a translational research perspective is that RBM has both human and murine content, with significant overlap in the covered analytes. So one can do a biomarker discovery study in a mouse preclinical model and then confirm the findings in clinical samples with the same vendor and technology platform. The NIH should probably fund a consortium project for companies to make an antibody to every protein, including post-translational modifications, in the proteome. These reagents could then become a commodity, and scientists/companies could build their intellectual property around what they discover using them. A shift toward use patents over composition of matter patents would occur. You hear anecdotally that one or another company is doing this, but I have not yet seen anything substantial launched. I think it has been happening in fits and starts, but unfortunately it has kind of stalled.
Cambridge
Healthtech Institute| Beyond
Genome | Bio-IT World
| Biomarker
World Congress | Digital
Healthcare & Productivity | Your Life Science Network Cambridge Healthtech
Institute | 250 First Avenue | Suite 300 |
Needham, MA 02494
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
