Lund University's Laurell on the Pros and Cons of Sandwich and Reverse-Phase Arrays

Name: Thomas Laurell

 

Title: Professor in Medical and Chemical Microsensors, Lund University

 

Professional background: 2002 – present, professor in medical and chemical microsensors, Lund University, Lund, Sweden.

 

Education: 1995 — PhD, Lund University, Lund, Sweden.

 

 

Thomas Laurell is one of eight researchers that joined together last year to found Create Health, a translational research center at Lund University in Sweden that integrates bioinformatics, proteomics, and nanotechnology in one environment. To date, Laurell’s work has focused on developing microfluidic and microarray platforms, and he has co-founded three microtechnology companies — Erysave, Picology, and ISET.

 

Apart from Create Health, Laurell is involved in several ongoing array development projects, including a partnership with Malmö University Hospital in Sweden that is creating a protein array tool that could be used to screen potential prostate cancer patients and determine a successful strategy for therapy. Part of that development work was highlighted in a recent paper in which Laurell’s team compared label-free reverse-phase assays with sandwich antibody assays and determined that the Malmö group would have greater detection capabilities using a sandwich assay, rather than the less-complicated reverse-phase assay [Järås K, et al. Reverse-phase versus sandwich antibody microarray, technical comparison from a clinical perspective. Analytical Chemistry. 2007 Aug 1;79(15):5817-25].

 

To learn more about the paper and the project, BioArray News spoke with Laurell this week.

 

 

Well, this is not specifically a Create Health project. I am running this project together with people in the clinical side in Malmö University Hospital. They have a very good and long track record in prostate cancer research. So we teamed up with the people a couple of years ago and started to work with microtechnology modes for diagnostics. They clearly saw some benefits of the things we are doing.

 

Specifically, [regarding] the reverse and the sandwich assay [study,] you see more and more papers coming out on reverse assays where the indication is that it could be a clinical platform due to its simplicity of running the assay. That is why we also wanted to check how our technology platform would perform in a reverse-mode concept.

 

Basically the platform we have is this sort of 3D, nanoporous surface that we array our antigens or antibodies on. These surfaces have proven to perform quite well as high-sensitivity surface[s] for spotted microarrays and protein microarrays. This is a development we did early on with Carl Borrebaeck’s group, within Create Health, in validating the quality of these surfaces. We are making and manufacturing these surfaces presently -- not in a high-throughput format; [but] it easily could be scaled up.

 

Nevertheless, the reason for doing this comparison was, as I said, the reverse [phase assay] is gaining a lot of attention and we started to check out how well we could do the detection on a reverse assay on these surfaces. We thought it could be an interesting way forward; just to spot the plasma sample straight onto the surface and then run the binding assay. That would be simpler.

 

But when we started doing this, we realized that we clearly had differences in detection level. After revising the protocols a number of times and not really seeing that we were doing anything differently or erroneous, it was clear that we were having a lower detection level as compared to when we ran the sandwich assay. So with that at hand, we thought it would be appropriate to highlight that as a fact and if you sit back and think about it a little bit, it’s quite evident that a reverse assay would not be as sensitive as a sandwich assay.

 

 

It’s clear that in some of the early papers there were reverse arrays [that] originally came from lysate arrays. So you were doing cell lysates and you took the intracellular material and you ran your assays on that. Looking at that situation, you commonly have a rather high expression level of the proteins you are looking for, even though it’s a limited material.

 

When you do that, reverse assays come out in a much better comparison. But if you are moving over to a peripheral body fluid like blood and you start to look for biomarkers there, the expression level of a biomarker in cancerous tissue may be expressed very high, but once that leaks out into the circulatory system you have a considerable dilution of the biomarker of interest.

 

That means that now when you do a reverse-phase assay, you are limited by the number of binders that you actually spot down on the surface. When you take a blood sample and you look for a low-abundance biomarker the number of binding sites on the spot that you put down are not very prominent.

 

In that sense, detection limits should be clearly lower than for the corresponding sandwich assay where you are utilizing high-binding density on the surface when you array the antibody down to the surface. In that case, you basically saturate the surface with a binder when you run a sandwich assay, whereas in the reverse assay, commonly if you look for the low-abundant biomarkers, you have relatively few binding spots on the sites.

 

So we come down to a detection level problem here, which is not always highlighted in terms of running reverse assays in serum or plasma unless you link an amplification technique to the straightforward fluorescence assay or if you move to more high-fidelity detection platforms like the Zeptosens platform.

 

Depending on what you want to do and where you want to go with the assay, it’s up for you to decide. But we want to direct this paper to indicate that if you are going to do protein microarrays you will need to consider this. I am pretty sure that you can’t just go on and array plasma straight down on a surface and hope to detect over-abundantly expressed interleukins, but you need to have protocols for doing that.

 

 

One situation I can think of for reverse assays is if you look into prostate cancer cases if you could acquire semen samples rather than blood samples. This is a biofluid that is much more linked to the disease location and also endocrinically linked to the disease, which means that you will have more of the biomarker in question in that fluid. So, if you have a biofluid confined to the disease region, that’s a situation where there could be a way forward.

 

But deriving a semen sample is not as easy as taking a blood sample. Also, it means that the clinic would have to implement new sample processing protocols, whereas the clinic normally today would like to see a serum sample coming in that fits into conventional scheme of blood sample handling and processing with either lysate assays or fluorescence-based assays.

 

So, some of this is written in the view of sort of slotting into the conventional labs that are out there. The clinics are not into developing completely new protocols for each and every disease. So reverse work is there and it will stay. But it should be pointed out that detection level is an issue, and it is the most common issue for people working with microarrays. If you can’t have good detection levels then you are basically out. It should be pointed out that the sandwich assay also confirms the actual binding of the antigen to the primary antibody, which inherently gives a quality control of the diagnostic readout.

 

Currently we are doing this as a development project with the university people, and the goal we have is to basically set up a small multiplex assay. We have been working on this [for] some time now, feeling confident with the detection levels that we are showing. We are moving on with a couple of other biomarkers that are very closely linked with PSA and the disease that have been shown in other studies to show a strong correlation.

 

So if you can run an assay on, let’s say, three or four biomarkers in parallel on a very small sample with the read-out quality we have currently, we are quite happy in terms of moving on now to a setting where we could start screening in population-based studies. These four markers would need to be validated on a much broader cohort that is well-characterized from a clinical point of view and a disease progression point of view. These are biobanks that our colleagues have access to and they have control over those biobanks.

 

The goal is mainly to select a good diagnostic pattern from these arrays to distinguish benign from malignant disease in prostate cancer patients. We believe that clearly there should be a way forward to make a small, diagnostic multiplex device that indicates whether you need to go for a prostatectomy or if medication would be sufficient or other treatment would be sufficient.

 

It’s a rather big step to actually undergo prostatectomy. Also, roughly 75 percent of people coming in diagnosed with two high PSA level are actually not malignant and the situation of making a decision to move on with prostate biopsy work which is painful for the patient and elaborate for the staff. We clearly have a problem in accuracy in disease diagnostics here.

 

 

I would say that it would probably be quicker than the US, but that doesn’t say very much. The current situation is as follows: This spring the Swedish Food and Drug Administration turned down a proposal of implementing PSA testing as a standard screen for all men over 40 in Sweden. It was turned down and the arguments against it referred to the inaccuracy in predicting a malignant disease.

 

So right now, based on only a single biomarker, there will be no screening activities going on. There will have to be studies over time, and the good thing is that longitudinal material is available in the biobanks that we have, where follow-ups on patients over a period of 30 years have been done. This means you can follow disease progress over time and look at the biomarker expression path. But again, this requires some extensive studies, so I don’t see anything changing in the next few years for sure.