Medical Device Link.

IN PERSON

Graham Lidgard, PhD, is senior vice president of research and development at Nanogen Inc. (San Diego). He can be reached at glidgard@nanogen.com.

IVD Technology: What have been the most significant advances in detection technologies over the past few years?

Graham Lidgard: For basic technologies, chemiluminescence and fluorescence detection have dominated what is going on in the market.

Because it operates against a dark background, chemiluminescence offers many benefits, including high sensitivity and low background. And with the photomultiplier tube (PMT) technology that has been available, detecting very small amounts of light has become very efficient.

The birth of chemiluminescence technology in diagnostics dates back to the Stuart Woodhead technology—acridinium-based chemiluminescence that started with Corning Medical and is now part of Bayer Diagnostics (Tarrytown, NY). It’s the basis of the ACS 180 and Centaur technology.

Other companies that have been active in this area are Abbott (Abbott Park, IL); Kodak, whose technology is now part of GE Healthcare (Piscataway, NJ); and Roche Diagnostics (Indianapolis), with its Elecsys systems. So, from an amino acid perspective, chemiluminescence has probably been the predominant technology.

In fluorescence, some of the similar rules for chemiluminescence apply. It’s a light-based technology. There’s a camera and fiber-optic technology present. The electronics industry, with its development of both lasers and light-emitting diodes, has provided tremendous background for instrumentation. And the basic chemistry technologies used really depend on the instrumentation to detect them.

Real-time polymerase chain reaction (PCR), end-point PCR, in situ hybridization, and now microarrays are all based on fluorescence technology. There are some other technologies that are being proposed, but they really don’t have the basis in experience and technology that is associated with fluorescence.

In terms of sensitivity, I don’t think any detection technology has outperformed chemiluminescence. That said, I think the technology has more or less reached a saturation point. I’m not sure there’s much more to be gained out of it. Then again, such things are always difficult to predict.

What are the latest trends in developing detection technologies? Is there much room for chemiluminescence technologies to develop?

People are looking for a high degree of multiplexing. With chemiluminescence, it is difficult to differentiate wavelengths and multiple targets. As soon as you try to differentiate on wavelength, you lose a tremendous amount of sensitivity.

However, with fluorescence, you do have the ability to look at many different wavelengths. Although quantum dots encounter some problems of their own, you’re able to create dyes with different absorptions and emissions. Using an array-based system, you can actually expand this by spatial dimensions. So, by having different positions on an array, you can increase the high degree of multiplex.

Everybody’s looking to be able to do more-complex multiplexing in the future—in areas such as cancer diagnosis with gene expression, pharmacogenomics with SNP detection, and infectious-disease testing. The diagnostics industry is moving from the era of asking whether a patient has flu or chlamydia or pneumonia, to asking what they have. When researchers and doctors start asking that question, they need a panel that’s based on what is likely to be in the sample. So, if it’s a respiratory sample, they’d like to have all the respiratory viruses present. People are realizing that this technology is now being enabled. Clinical researchers and diagnostics companies are starting to move in that direction.

With this desire for multiplex in mind, what challenges do IVD manufacturers encounter when designing and developing detection technologies?

I think the biggest challenge is looking at the whole, sample to answer, as a system. A lot of early development work goes into producing detection technologies without realizing that an instrument has turned into a $300,000 piece of equipment with a cost per detection of $50 or $100.

It’s important to start development by looking at the whole system in terms of sample prep. How much target is available for detection? How do you process this to a point where it can be detected? Sensitivity has always been a driving force in instrument design, but sometimes sensitivity isn’t the answer.

For instance, everybody in the market has experienced a tremendous false-positive rate when testing for diseases like Chlamydia trachomatis and gonorrhea. A sample may have a target of 106 or 107 organisms. A high degree of sensitivity isn’t needed for this. What is needed is specificity.

You have to look at the system as a complete system and determine its function in relation to the test. The detection drives throughput, specificity, and sensitivity. The big challenge for manufacturers is to bring all of these pieces together.

There are a lot of systems that use unique detection methods. For example, a glass-slide microarray with a fluorescence detector. A lot of software algorithms are needed to analyze the spatial distribution of the array. But there’s also a lot of work that needs to be done to get to this point—preparing the sample, doing amplifications, preparing the reagents. A challenge that manufacturers face is to build sufficient detection technology for the job, but not to overengineer it. And not to make it so expensive that the system is not viable in the market.

Finding a Balance

How can IVD manufacturers go about making their detection technologies both more sensitive and more specific?

The first molecular biology workstation by Nanogen was a research-based instrument. We did DNA amplification in addressing the target to a particular location on the microarray. Then we conducted a hybridization and fluorescence detection. We built a system that had high-performance lasers. Also, it demonstrated very accurate precision movement within the system and about three levels of sensitivity.

We found that with the chemistry and the way we were able to optimize the chemistry, we only ever used the lowest level of sensitivity in the system. So, for the next-generation instrument, we’ve switched the detection technology to a high-performance camera system, which is similar to the cameras that are used on optical space telescopes.

With this, we can achieve the same, or even better, sensitivity in a few seconds compared with the five or 10 minutes needed with the laser-based system. We learned that it’s not always the most sophisticated technology that’s the best. Instead, it’s about tuning the technology to the application.

How can a company achieve this happy medium—a dependable detection technology that is sensitive and specific, but also stays within cost confines?

The rule I’ve applied during the 30 years that I’ve been developing diagnostic systems is what I call the “30-30-30 ratio”—30% of the development process is chemistry, 30% is detection instrumentation technology, and 30% is software. And the other 10% is usually luck.

A lot of companies will try, at the last minute, to solve all of their problems with software. They will reach a point at which the hardware and chemistry are basically sound, and they’ll use software to fix sophisticated imaging algorithms or correction factors for fluorescence—any number of things either across an array or within a solution. But you really have to maintain a balance. If you try to solve all of your development problems using one part of the instrument, you can end up with a system that’s not only difficult to manage during development, but also in the field.

This can also create problems during validation, where you have to be able to artificially induce problems and then show you’ve solved them. So, sometimes you have to go back to your hardware and be prepared to throw out one design.

In the end, it’s the chemistry that delivers a signal. The hardware looks at that signal and interprets it into a quantitative or qualitative response. The balance between the three can involve very sophisticated chemistry, and it can be very expensive to build equipment around.

In considering sensitivity, I always go back to a 1976 paper by Tomas Hirschfeld out of the Los Alamos National Laboratory (Los Alamos, NM), perhaps the first demonstration of the fluorescent detection of a single molecule. The researchers coupled a polymer of fluorescein to an antibody, shot a laser through a microscope into the solution, and then, when the molecule passed under the laser, the laser was so strong that it produced instant photobleaching.

The problem was that you had to have 100,000–200,000 molecules in solution for any one molecule to pass in front of the laser in any reasonable time of measurement. If you have a single molecule, I think they calculated that it could be about three months before you actually saw it. So, the absolute sensitivity of the technology is not the total answer. Rather, it’s the concentration you can detect in the primary sample when you’re all done.

Sometimes people will say that they can measure one to five copies in their PCR. However, if you can only get one microliter of your primary sample into your PCR, then you’ve got 1000 times less sensitivity.

It’s a matter of looking at the whole system. If you discovered a way to concentrate a milliliter of solution into one microliter, and to put that into the amplification, then you’ve got more sensitivity than anybody else without changing your PCR.

How have developments in molecular diagnostics affected the development of detection technologies? And what effect have other fields such as nanotechnology and pharmacogenomics had?

Molecular technology is still in the early adoption phase of the market. There are a number of high-volume tests, for diseases such as HIV, HCV, chlamydia, and gonorrhea, that have essentially reached their peak in the market, but the technology as a whole is still growing.

What we’re seeing now is that the capability for multiplexing and molecular technology is really emerging from the cancer research market, from pharmacogenomics. Being able to measure each variant separately in a system would be cost-prohibitive and difficult to do. You could be talking about 20–30 targets per gene. And if you had to perform each one of these with an individual PCR, the cost and time and throughput required gets burdensome.

One of the interesting features of the Nanogen array system is that you’re able to put multiple patient samples onto a single chip. Other instruments use one sample per array, and that has presented a cost limitation. The technology produced by Affymetrix Inc. (Santa Clara, CA) is very elegant and can detect many things, but it’s still extremely expensive. If you are using a glass-printed array, the cost of the printing and the quality control of the printing of an individual array for each patient sample can get expensive. By combining an array with a multipatient system, we have been able to dramatically bring down costs. Also, our system provides an open-platform system through which customers can develop their own assays. The ability to develop smaller-volume assays can also lower costs.

As far as detection in nanotechnology is concerned, we’re still waiting to see. The technology for manipulating nucleic acid targets is so good that you can solve a lot of detection sensitivity problems just with amplification. The higher-sensitivity technology that’s been promised with quantum dots or fluorescent beads really hasn’t happened yet. These new technologies have their own problems. We know a lot about chemical fluorescent dyes—how they bleach and their quantum efficiency, for instance.

We’ll be able to solve these problems but each brings its own complexities. For example, with some of the quantum dot technology, as you expose it to fluorescent signals, you get an increase in fluorescence. It’s not a steady state.

Even the small size of some of these beads is too large—1 µm compared with the 50-µm pad in our detection system. So, we have to get to smaller particle sizes for nanotechnology to be able to complement an array-based system.

Evolving Technologies

What sort of detection technologies has Nanogen developed?

We signed an agreement to merge with Epoch Biosciences (Bothell, WA) in September 2004. Epoch has developed some real-time detection technologies that are different from the existing TaqMan-based systems. They’ve developed probes that are not cleaved and are able to operate in melting scenarios. So, not only can you multiplex based on fluorescence, but you can also multiplex based on the melting-curve detection of the target.

They’ve also developed nice nonstandard nucleotides, which can overcome some of the complexities of DNA targets. For example, a lot of guanine sequences in a row can make it hard for amplification technologies to pass through those sequences. The nucleotides developed by Epoch eliminate that problem and allow PCR and other amplification systems to operate more efficiently.

We’re also applying these to microarrays. The company’s nonstandard nucleotides can eliminate the problem of variance in a target. The group has produced a number of papers on enterovirus in collaboration with ARUP Laboratories (Salt Lake City). Enterovirus consists of a whole range of organisms with sequence variations. As a result, being able to design and develop probes that can detect the entire sequence is complex.

To address this, Epoch has produced a technology called the minor groove binder (MGB). This allows a reduction in the size of the DNA probe so that researchers can conserve sequences. Then, the nonstandard nucleotides can be added to reduce the effect of variation. Even though this is not the direct cause of the signal, we consider it a detection technology.

A lot of that technology has been licensed out. The MGB TaqMan probes by Applied Biosystems (Foster City, CA), for example, are probably one of the most widely used technologies in the real-time research market.

For microarrays, we’re into our second generation. These products use a smart chip. This is a complementary metal oxide semiconductor, or CMOS, which contains all of the electronics that control the pad electronics. The instrument communicates to the 400 pad through 12 connections to the chip.

Using this technology, we’re able to electronically change the charge at a 50-µm location on the chip and cause DNA to migrate to that position very quickly. As a result, we speed up hybridization. What would take 16 hours in overnight-type incubations, we can do in less than a couple of minutes.

The whole array technology is moving toward a more rapid process. And we’re able to put multiple samples onto the chip and do complex detections—20 or 30 SNP detections for 50 or 60 samples—in a single run.

The expansion in the pharmacogenetics field is going to create a demand for this type of technology. And our array is an open array; the DNA is not predetermined. So, it’s the same chip for all of the systems that get developed, whether by the customer or by us.

Does Nanogen have any other partnerships or strategic alliances with companies to develop new detection technologies?

We’re always talking to companies that are working on new detection technologies. Epoch has a tremendous background in nucleic acid chemistry that’s able to go onto most of our systems.

As far as outside collaborations go, we’re concentrating more on targets. There are a number of researchers and companies working on specific SNPs, specific DNA sequences, and viral detection. One example is a collaboration we have with a company in Finland called Jurilab, which has a unique population database that goes back tens of years. It records instances of heart disease, high blood pressure, and other conditions. We’re working with them to analyze the genome of those samples to find sequences that correlate with prediction of developing those conditions. Currently, we’re cooperating on a type 2 diabetes program.

We also have a number of collaborations with reference labs and clinical labs for developing assays that will fit onto our systems. But as far as detection technologies go, we haven’t seen anything yet that would cause us to significantly change direction. We are working with a lot of labs on new amplification systems, but they are in the early grant stages of research.

What sorts of new technologies would cause Nanogen to change direction?

One characteristic of technologies that we’d be interested in is a broad application. A number of amplification technologies have been developed, but they work in a very small area of the nucleic acid. They have problems with certain sequences and certain applications.

Also, speed. How quickly can the detection be completed? If you look at the history of diagnostics, from chemistry enzymes to immunoassays, we’ve gone from two- or three-day tests to ones that can be conducted in seven to 10 minutes. I think that molecular diagnostics is going to continue this trend.

Another important factor is cost. If you have to put an expensive piece of instrumentation around the detection technology, that can serve as a deterrent. Building an instrumentation system in the diagnostics industry is very expensive. The simpler, more robust you can make it, the stronger an asset it will be.

The Future of Detection

What future challenges will IVD manufacturers face in developing detection technologies?

Certainly, there is a strong market desire to have a point-of-care type of nucleic acid technology, in which the sensitivity of nucleic acid detection can be joined with the speed and convenience of immunoassay detection. I’m thinking, for instance, of rapid-care lateral-flow technologies.

To be able to overcome this challenge, a test must be able to produce high sensitivity in a very short period of time. And this has to include the sample through to the answer. The detection technology won’t seem impressive if it takes three or four hours of chemistry and labor to get to it. An integrated system in which sample to answer can be completed in under an hour is ideal. A test performed in 15 minutes is the holy grail, but I think there’s still a significant market for the performance and simplicity of a test that can be performed in under an hour.

It also comes down to keeping the instrumentation cost low. The challenge, then, is not so much the detection technology, but rather integrating it into a system that meets the speed and cost needs of the market.

One example of this is the TDx system from Abbott, based on a fluorescence polarization immunoassay (FPIA) technology. The technology was written off by most of the industry in terms of applicability, sensitivity, and complexity. In the 1980s, Abbott came out with a very simple machine, but used the FPIA technology. And this system has pretty much dominated the drug-testing market for a number of years. So, it’s not necessarily the elegance of the detection part of the technology, but rather how well all the pieces fit together.

Looking at detection technologies, what challenges do manufacturers face when developing point-of-care instruments and devices?

One of the divisions that Nanogen acquired, Spectral Diagnostics Inc. (Toronto), manufactures and sells rapid-care diagnostics for the cardiac market. We currently have an NT-ProBNP product in development for congestive heart failure. Our interest in this business was to learn about and build a presence in the market, and to be able to then apply our technologies to more-rapid and more-sensitive detection in the rapid-care area.

A lot of the newer targets are being identified at much lower concentrations. Detection for future rapid-care products will need to be in the 1–10 pg/ml range for proteins. So, the question becomes: How do you drive that? There’s a lot of fluorescence technology being used. People are also looking at quantum-dot and bead fluorescence technologies as well as magnetic-particle bead technology. Many technologies are in the early phase of development and feasibility evaluation.

But again, what’s important is being able to integrate a highly efficient technology for low concentrations in a low-cost rapid system. The market is very restrictive in terms of the expensive equipment it will tolerate. If you’re doing only a few tests a day on the system, it can’t be a high-complexity instrument that requires a lot of service and support.

What new detection trends can we expect to see this year and in the future?

I think you’re going to see a lot of nanotechnology feasibility—things like nanowires, nanodetection, and nanobeads. That seems to be a big research interest right now. The challenge will be converting these technologies into practical solutions. My experience is that, from the first time a new technology in our industry is published, it can be five years before anybody has a practical application for it. And it can easily be 10 years before it actually enters the market in an economic format.

As a result, I think we’re going to see a lot of early feasibility studies for these new technologies. In particular, there’s a tremendous amount of funding going on in the biodefense industry around high-sensitivity rapid detection. It will be interesting to see which of these technologies meets users’ criteria and becomes a dominant technology.