Tag Archives: biomarkers

From Immuno-PCR to Peptoids: Why Great Ideas Sometimes Aren’t

From the inbox last month:

Dear Dr. Dove:

Just curious. Do you really think Kodadek’s Cell paper is a good piece of work? The response of the science community to this seemingly amazing news is silence. This usually means that the results of the paper are odd.

Wang fan

Thanks for your note, Wang, and I’m sorry it’s taken me a month to get back to you. This question touches on some much broader issues in science and science journalism, and I wanted to take the time to put together a proper blog post about it. For readers who don’t know the background, here is my original post about Kodadek’s work, in which I unabashedly raved about a novel technique for discovering new disease biomarkers.

I stand by that assessment. The researchers performed an extremely clever experiment, it apparently worked, and they followed up on it pretty carefully. That’s what science is supposed to be about.

So why isn’t everyone using Kodadek’s strategy to find a slew of new biomarkers already? I can think of a few possibilities. One, of course, is that the technique might be a lot harder to perform than the paper suggested. Perhaps it yields inconsistent results for different diseases. Perhaps there just aren’t good biomarkers for some of the diseases we want to study. Or perhaps people are all over this technique, and we just haven’t seen the papers yet – it hasn’t quite been two years, and these studies would take some time. It’s also possible that the method just isn’t as useful as it first seemed. These sorts of problems trip up new ideas a lot more often than even most scientists realize.

Consider, for example, the tangled tale of another biomarker detection scheme: immuno-PCR. Developed by Charles Cantor’s group way back in 1992, immuno-PCR uses the highly effective signal amplification of polymerase chain reaction (PCR) to detect proteins. The technique was supposed to solve one of the biggest problems in biochemistry at the time, which was identifying and quantifying proteins that are present in vanishingly small quantities in a sample, which certainly describes many promising disease biomarkers. Immuno-PCR was much more sensitive than the best available competitor, a technique called enzyme-linked immunosorbent assay (ELISA), and more sensitive protein assays really sounded like just the thing. The new method was poised for greatness.

But it flopped, or at least went to sleep for awhile. In 2008, I wrote a feature article for Bioscience Technology magazine about protein detection methods, and pretty much everyone I talked to agreed that immuno-PCR was a disappointment. Indeed, several companies were trying to develop a replacement that would work better. It turned out that immuno-PCR was just too hard to do properly, so most users gave up on it.

Similar problems befell David Ward and his colleagues at Yale University, who developed an equally clever protein detection system called the rolling circle immunoassay in 2000. Instead of PCR, Ward’s assay relied on a type of amplification called rolling circle replication, which many viruses use. This elegant new technique blew the doors off ELISA; the investigators could detect proteins at single-molecule sensitivity. Rolling circles were supposed to succeed where immuno-PCR had failed, but instead they also landed in the bin.

Some of the techniques developed since then have fared slightly better. A company called Nanosphere, building on work by Chad Mirkin and colleagues at Northwestern University, now has FDA approval for high-sensitivity diagnostic tests based on a nanoparticle detection system that uses DNA “barcodes” to identify individual proteins in a sample. Nanosphere’s technology apparently offers the sensitivity of techniques like immuno-PCR and rolling circles, without the same technical headaches.

While successive groups of scientists were working on these assays, though, advances in completely different protein analysis techniques made antibody-based detection somewhat less relevant. The past decade has seen astonishing advances in protein mass spectrometry, which allows researchers to identify and quantify proteins without having to make antibodies against them first. Why bother with DNA-bound immunological probes when you can simply feed your sample into a box and read a list of the proteins in it on your computer screen?

Meanwhile, we’ve learned more about protein-based assays, especially in medical testing, and it turns out that greater sensitivity isn’t always a good thing. Barcoded nanoparticles can detect previously undetectable levels of prostate-specific antigen (PSA), for example, but a growing body of evidence suggests that PSA testing does more harm than good. Making a bad test more sensitive only makes it a worse test.

Anyone who’s been a science journalist for more than a few years has probably collected a whole slew of similar stories: results that just didn’t pan out. That’s why I always try to discuss the limitations of a new paper as frankly as possible, even in the midst of an unabashed rave.

So what happened to peptoids? It’s too early to tell. Even if the technique proves troublesome, it represents a fresh approach to a question that currently looks important: what bloodstream biomarkers can we measure to diagnose and monitor chronic diseases such as Alzheimer’s and cancer? 18 months ago, peptoids looked like they might be a really cool strategy for exploring that topic. Maybe they still are.

References

1: Sano T, Smith CL, Cantor CR, “Immuno-PCR: very sensitive antigen detection by means of specific antibody-DNA conjugates,” Science, 1992 Oct 2;258(5079):120-2.

2. Schweitzer B, et al., “Immunoassays with rolling circle DNA amplification: a versatile platform for ultrasensitive antigen detection,” Proc Natl Acad Sci U S A. 2000 Aug 29;97(18):10113-9

3. Hill HD, Mirkin CA, “The bio-barcode assay for the detection of protein and nucleic acid targets using DTT-induced ligand exchange,” Nat Protoc. 2006;1(1):324-36

4. Thaxton CS, et al., “Nanoparticle-based bio-barcode assay redefines “undetectable” PSA and biochemical recurrence after radical prostatectomy,” Proc Natl Acad Sci U S A. 2009 Nov 3;106(44):18437-42. Epub 2009 Oct 19

The Peptoids are Coming

It’s too bad we’ve abused the word “innovative” so horribly, because every now and then I come across a bit of new research that really deserves the title. This week it happened again, with a paper that’s scheduled to appear in Cell right about now. On its surface, it doesn’t sound especially promising: the researchers, led by Muralidhar Reddy and Thomas Kodadek at Scripps Florida, claim to have found some potential biomarkers for Alzheimer’s disease. Announcements like that flood my inbox, and seldom amount to much. I think this one is a bit different.

There are three amazing things about the new study, but before I explain them I’ll put the finding into context. A “biomarker” is anything we can measure that distinguishes people with a disease from people without a disease. If you’ve ever been tested for a viral infection, such as HIV, hepatitis B, or H1N1 influenza, you’ve experienced a biomarker assay. The first-line tests for those and many more viruses are actually looking for antiviral antibodies in your blood. Because they’re so easily accessible and detectable, antibodies are the markers of choice in biomedicine, and antibodies against a virus are a pretty strong indicator that you’ve either been exposed to it or vaccinated against it.

For many diseases, though, scientists have had a very hard time finding biomarkers. Pathologies such as Alzheimer’s disease and cancer certainly change the mixture of antigens in the body, and everything we know about immunity says that there should be corresponding changes in antibodies. In theory, we should be able to detect those changes and use them as biomarkers.

Unfortunately, there’s a chicken-and-egg problem. If you don’t know what the antigen is, it’s tough to fish out antibodies against it. Viruses are the low-hanging fruit, because all we have to do is isolate the viral capsid and use it as a probe for the antibodies we want. Researchers have also had some success with autoimmune diseases, such as lupus, where the pathology of the disease provided some hints about what antigens the immune system might be attacking. For many other non-infectious diseases, including Alzheimer’s, we’re groping in the dark.

Reddy et al. decided to use an entirely novel approach to this problem. They started by using a technique called combinatorial chemistry, which allows them to synthesize a huge number of different molecules from common building blocks. It’s a standard strategy in the drug industry, but the Scripps team’s approach had a twist: instead of conventional drug structures, they built their molecules from artificial amino acid-like building blocks, yielding short sequences called peptoids. That produced a library of thousands of structures that look more or less like small protein fragments, but that aren’t quite the same as regular biological proteins. In essence, it’s a library of different shapes representing the universe of possible antigens, without bias.

Next, the investigators sampled blood from mice with or without an experimentally-induced autoimmune disease called EAE. The peptoids identified a few antibodies that differed between the diseased and control animals, and those antibodies held up very well as biomarkers in a subsequent test on blinded samples. Apparently, peptoid libraries can be used to discover novel, reliable diagnostic assays for a disease without being given any initial information about what the antigens are. It paves the way for a whole new approach to hunting biomarkers. That’s the first amazing thing.

Having proved the concept, Reddy and his colleagues moved to testing blood from Alzheimer’s disease patients and healthy controls. Again, they found some antibodies that can distinguish the two groups reliably. So this brand-new approach has immediately provided a set of promising leads for Alzheimer’s biomarkers, a field that has been struggling for years. That’s the second amazing thing.

The third amazing thing was the press release Cell put out to accompany this new paper. It would have been easy, and almost standard procedure, to hype these findings through the roof: “Researchers develop reliable blood test for Alzheimer’s disease,” or “New method is a game-changer for diagnosis.” Instead, I got the barely-worth-a-click headline “A blood test for Alzheimer’s disease?” Yes, with a question mark*.

Under the headline, the text is just as cautious:

Kodadek says they have since extended the test to more patients and it appears to be holding up well. Nevertheless, development of a clinically useful test will depend on further validation. It’s possible that the test might not work as reliably well in a collection of patients representing different ethnic groups or different forms of dementia, he cautions. They’ll also need to transition their peptoid technology to a simpler platform better suited for use outside of a research laboratory.

It’s not entirely clear whether an early test for Alzheimer’s disease would be broadly useful today given that there aren’t any real treatment options, he added. Such a test might initially be most useful to pharmaceutical companies, by allowing them to better identify patients with early Alzheimer’s for enrollment into clinical trials.

I’ll add one more cautionary note. We don’t know whether the Kodadek lab is detecting a useful early biomarker for Alzheimer’s disease. They might just be seeing the antigen changes that occur later in the disease, when physicians could just diagnose it from the symptoms. Nonetheless, it’s the coolest new technique I’ve seen from the biomarker field in a long time.

* Of course, I also received a somewhat more hype-enriched press release from Scripps, but even it isn’t so bad once you get past the headline.

Reddy, M. M. et al., Cell 144, 132-142, January 7, 2011, DOI 10.1016/j.cell.2010.11.054.

Probing the Proteome

My most recent piece for Science/AAAS is now online. I talked to several researchers who are using very cool techniques to find new biomarkers for diseases, and it was a fun piece to write. It appeared in the “AAAS Business Office” section of the magazine, which means it ran on advertiser-supported pages. There’s no “advertorial” content (I don’t do that sort of thing), but being advertiser-supported does mean that it’s free to the public. Follow the link and enjoy.