NAME: Gareth Gerrard
POSITION: Research associate, Adult Leukaemia MRD Unit, Imperial College London.
BACKGROUND: Biomedical scientist, Royal Free Hampstead NHS trust.
Measuring BCR-ABL1 transcript levels using quantitative real-time PCR has become an important tool for monitoring minimal residual disease in patients with chronic myeloid leukemia.
As such, researchers are currently investigating ways to optimize such assays by increasing their sensitivity and throughput while reducing cost. One such group, in the Leukaemia Minimal Residual Disease Unit and Centre for Haematology at Imperial College London's Hammersmith Hospital, has been at the fore of these efforts, having developed and implemented a BCR-ABL1 qPCR assay that uses ABL1 transcripts as an endogenous control.
Although the assay has worked well in practice, in an effort to further reduce cost and test turnaround time researchers in the unit have recently attempted to turn the two-plate qPCR assay into a single-plate, duplex assay, while maintaining the sensitivity of the original assay.
The researchers documented their effort thus far in an article published this month in the American Journal of Hematology. This week, first author Gareth Gerrard took a few moments to discuss the team's work with PCR Insider. Following is an edited transcript of that interview.
The [BCR-ABL] assay that you were using in your clinic, prior to the new assay, was that a commercial assay, or was it developed in house?
It was very much developed in house. [At] this particular lab at Hammersmith [we are] involved in looking at minimal residual disease in chronic myeloid leukemia. This is characterized by BCR-ABL1, the Philadelphia chromosome, which is a fusion oncogene leading to a constitutively activated tyrosine kinase. This is very useful as a marker in minimal residual disease, because it's absolutely unique to the tumor cells. Basically, it shouldn't be found in normal cells. Also, the fusion genes are very good targets, usually, for quantitative PCR assays, because you have this aberrant transcript being produced with one gene [fused] to another gene, so you just produce a forward primer in one gene, and a reverse [primer] and a probe in the other gene, and in theory you should be able to quantify it very accurately.
I wasn't involved in the development of the original assay – it was sort of a legacy assay that came to us from Nick Cross, [currently at the University of Southampton School of Medicine and Wessex Genetics Reference Laboratory] and Jaspal Kaeda, [currently at Charité-Universitätsmedizin Berlin], who [were] here at Hammersmith until about 10 years ago. They were also involved in the Europe Against Cancer initiative, which … set about the task of … trying to standardize all of these fusion gene assays that were being used, not just in CML, but also acute lymphoblastic leukemia.
So they came up with these TaqMan assays, which have been sort of standardized and tested across many labs in Europe. They're sort of standardized primers and probes; standardized dual-labeled TaqMan with FAM-TAMRA [FRET probes]. I think they were all designed using [Applied Biosystems] Primer Express software.
And these were being used in labs across Europe?
Yes. And the only thing we did differently was the plasmid we used for the standard curve, and again, I think it was used before in a more primitive assay. So instead of the G6PD insert in exon 3; which unfortunately is exactly where the [Europe Against Cancer, or EAC] ABL primer would have sat … so … Hammersmith, Jaspal, [and others] who [were] here before designed a separate ABL primer-probe, which was a bit further down, sitting more at the end of exon 3 or beginning of exon 4. So that's really the way we [deviated] from the EAC protocol, simply because of the materials we had here. But in all other respects, it was completely the same.
And that was run in two separate plates — one plate for the BCR-ABL, and then a separate plate for the ABL. The way we run the assay is that the ABL is the endogenous control, with obviously the BCR-ABL as the gene of interest. So all other things being equal, the plasmid should give us a nice one-to-one expression ratio. That's sort of the adopted standard now across Europe, although some people still use variations of the endogenous control gene. But generally, the recommended one [is ABL]. We just produced some UK-level guidelines, which are being published later this month or next, which sort of … recommend this protocol for measuring BCR-ABL.
But you needed to change the assay from two plates to one to save time and money.
Right, that was basically it. We're a very high-throughput lab. I was using duplex assays for research … and I was using very small amounts of sample and duplexing and multiplexing where I could. And I thought, 'Well, why can't we do this for BCR-ABL?' But it actually turned out to be more difficult than simply mixing the two together and changing the fluorophore. It was a very standardized assay, and we wanted to make the minimum number of changes. And it took me a long time to validate the assay — the various steps beforehand, the primer-probe matrices, trying to find the right level of probe level and primer level, the volume, et cetera.
But we were eventually able to make very small changes to the original assay. The key to this was the master mix. When I first started this I was trying to use the ABI universal master mix. One of the things you can normally do with endpoint PCR is, rather than using it at 1X, just use it a bit stronger, 1.1 or 1.2X. But that didn't work in the real-time setting. But then we brought in this Gene-Expression master mix (also from ABI), which initially worked really well. But recently I think ABI made some changes to the master mix, and just as the paper was being published, we've found that something has changed the kinetics of the assay ever so slightly, which we found sort of troubling. We are working with ABI right now to resolve this. We tried some other master mixes, particularly from Qiagen, and they worked, but the ABI one worked best, and gave the clearest signal and the smallest noise-to-signal ratio.
Do you know why one master mix might work better than another?
Well, Qiagen would normally tend to optimize their things for Qiagen machines; and ABI would tend to optimize for ABI machines. And our workhorse analyzer in the lab is the ABI 7900 HT. Even though the Qiagen reagent worked perfectly well on that, I think maybe the ROX level was a little low … But I suspect if we hadn't compared the two I would have been perfectly happy with the Qiagen. The Gene-Expression master mix worked best, and allowed us to move forward with the project and get it validated. But as I said, ABI is looking into why the two most recent batches changed a bit.
What other changes did you need to make to the assay?
It's funny, because this was never meant to be the endpoint; it was meant to be a first step, and we were going to look into using other technologies, like [minor groove binder] probes, which are a bit shorter and maybe a bit more specific; and [look into] maybe introducing a third color or something like that, because again, we're only using a single endogenous control gene. And usually when you're looking at gene expression, it's best to use two or more. We're actually working on [an assay with] the MGB probes right now, and working to maybe make it EAC-compatible, because the majority of the labs in Europe use the EAC-ratified primer-probe sets. They also often use commercially produced plasmid for their standard curves, which is made by Ipsogen.
But the main thing here is shortening the probe length, because what we found with the duplex assay, and I sort of [confess to] this in the paper, is that for some reason there is some sort of reaction that makes it ever-so-slightly more sensitive; but also we get a little bit more background, and I wonder if it has something to do with probe-probe interactions. So one thing we're thinking is that shortening the probes and using MGB or locked nucleotide technology would give us less of a target for probe-probe interactions. And another thought is to see if maybe running this in so-called fast mode, using another ABI fast advanced master mix, will maybe help keep the background lower, without loss of sensitivity, because obviously we're a minimal residual disease laboratory, and we're interested in getting this down to one-transcript sensitivity. Even though we can't really quantify accurately below, say, 10 transcripts, which is the limit of the technology; we can see positivity down to one transcript, and we are interested in keeping that sensitivity.
So at the moment I'm working on that to see if I can get everything running, because as it stands, the duplex assay that we have is only of limited interest, because we're the only people that use that ABL primer-probe. But [it's looking hopeful] that we can get the EAC-compatible one up and running.
You're the only ones that use the ABL primer-probe?
Yes, the so-called Hammersmith ABL probe set. There may be other people out there using it; I think there are a couple of WT1 assays out there that use this particular probe set, because some of the people at those labs were probably involved with this project at some point. But as far as I know, we're the only people doing BCR-ABL quantification using this particular ABL primer-probe set. In a way, this was meant to be a proof of concept to see if it would work on our assay, and then see if we could expand it to an assay that more people could use, as I know there are quite a few people in the UK that are interested in this assay and are sort of keeping an eye on our progress.
So the end goal of this is to develop this duplex assay under the standards of this EAC initiative. You're considering other protocols and master mixes?
Yes, absolutely. I mean we definitely went live with this assay, and it became our standard and was very popular in the lab here. But we're processing between 30 and 50 samples a day, and trying to get our turnaround to less than 10 days from receipt of sample to reporting. We also like the idea that by performing BCR-ABL and ABL in the same plate, we will increase our accuracy, as well, by sort of cutting down on that inter-run variation that you might get. And I think the paper showed that.
What are your next immediate steps for this?
The other thing now is actually opening this up to other targets. We're in the process of developing a WT1 duplex assay, with GUSB [beta-glucuronidase] as the endogenous control. And that works very well. And again, that's not using MGB, just the standard dual-labeled probe. It seems to work better when you have a non-fluorescent quencher, and that's something we're playing with too. We're not locked into buying from ABI, but we do use a lot of ABI products because that's sort of the history of the lab. If other people were setting up from scratch, then I would probably advise them into looking into non-fluorescent quenchers, because it's so much easier — if you get two colors, then you can go for a third color. Particularly if you're just doing general gene expression, [and with the BCR-ABL], then you want to get as many endogenous controls as possible.