NAME: Toumy Guettouche
POSITION: Director, new genome technology assessment and implementation, Hussman Institute for Human Genomics; and director, Oncogenomics Core Facility, Sylvester Comprehensive Cancer Center, University of Miami School of Medicine
Samples stored in tissue biorepositories contain a vast amount of phenotypic, histological, and pathological data. Unfortunately, molecular analyses of these samples — the majority of which are formalin-fixed, paraffin-embedded tissue blocks — have proved daunting due to the challenges associated with extracting high-quality genetic material.
For the most part, extracting RNA and DNA from FFPE samples is a manual process that is laborious, time-consuming, and lacking quality control standards, which causes a major bottleneck in front of downstream genomics studies such as real-time PCR or next-generation sequencing.
The John T. MacDonald Foundation of Human Genetics at the University of Miami School of Medicine is attempting to address this issue by marketing a new service for extracting RNA and DNA from FFPE samples in its CLIA-certified laboratory. The service, which is set to kick off in coming weeks, uses a variety of extraction and quality-control methods, the centerpiece of which will be the recently launched automated Tissue Sample Preparation System from Siemens.
Toumy Guettouche, director of research and development for the new service, recently spoke with PCR Insider about the new service. Following is an edited transcript of that interview.
Why did the Department of Human Genetics at the University Miami decide to offer FFPE nucleic acid extraction services, and how long has it been offering them?
We haven't actually offered it yet, because we're still finishing up some validation studies. But we're almost done; we are maybe a week away from completing everything we want to do to validate the extraction. In a CLIA lab, you have to do a lot more validation than in a research setting. Initially we intended to offer the extraction service in a research setting, but it quickly became clear that it would be very beneficial to offer that service in a CLIA setting. A lot of the customers that are looking at FFPE sample extraction come from the clinical side, and we wouldn't be able to help them if we ran this in a research setting.
But we are about to launch it. We've extracted more than 1,000 samples on the system [from multiple tissue types], and are happy with the results we've gotten, both for RNA and DNA.
You have obtained CLIA certification for this?
We have a CLIA lab, and are running the service within that lab. We are validating the instrument and the reagents in our lab, and do certain tests to make sure that the instrument is up to clinical standards. Validation studies on samples include looking at cross-contamination, consistency of the extraction yield, and quality of the extracted nucleic acids.
[Editor's note: The Siemens Tissue Sample Preparation system is Class I exempt in the US and is listed with the US Food and Drug Administration as an in vitro diagnostic device, according to a company spokesperson].
Why is there such a pressing need for this kind of service?
There are hundreds of millions of samples in tissue banks or biorepositories around the world. Most of them are in the form of FFPE tissue. Those samples are typically very well-characterized, so there is histological and pathological information about them; they have follow-up data. So these are very valuable samples that can be utilized for many different things, for example, biomarker discovery, clinical trials, follow-up studies, prospective studies, and things like that.
In the past, the difficulty of extracting nucleic acids from FFPE samples and the poor quality of extracted material have scared people away from using them. It's difficult to get good quality RNA and DNA out of these sections and extracted nucleic acids might look like they're highly degraded.
Another problem is that you have to use chemicals that are flammable, like xylene – most of the kits use some kind of solvent to get rid of the paraffin. There are some companies that claim they don't need solvents anymore — for example mineral oil has been used. Whether the quality of the extracted material is great, I don't know, because we haven’t tested them.
The problem with having to use solvents is that it's basically impossible to put them on a robotic platform. Those steps were typically done manually in a chemical hood, which takes several hours and is labor-intensive. So FFPE extraction therefore wasn't really amenable to high throughput.
But the need to look at FFPEs was there. There is a huge amount of information, a goldmine, hidden in these biorepositories. One reason why accessing the information hidden in these samples is difficult is because extraction was always low throughput, and typically the results of the extractions were poor.
There are also a lot of myths about FFPE tissue out there that aren't really true. For example, one myth is that old blocks don't work very well. We've found that the age of the blocks has very little to do with the quality of the nucleic acid that you extract. It seems much more dependent on how the formalin fixation was originally done. Let's say you take a piece of tissue in the operating room and then formalin fix it and store it. One really important factor in the quality of the nucleic acid one is able to extract is how long the piece of tissue was lying around before it was fixed. And then what type of formalin was used … and how long did they fix the tissue before embedding? Those are really key factors that probably play at least as important a role as the actual extraction method of the nucleic acid.
One reason I think old blocks don't work as well [is that] people used to use unbuffered formalin … and that leads to a lot more damaged DNA and RNA, more crosslinking.
Another myth is that DNA and RNA are highly fragmented. For RNA, that's certainly true, but if you look at traces ... you can actually see longer fragments, even more than 1,000 bases. Again, one reason I think you see so much degradation is the chemicals, but also the time it takes for the fixative to go through the tissue piece and stop the nucleases.
In the case of DNA we actually see some differences between different kits. Some kits seem to yield more fragmented DNA than others. And here the Siemens system was really great because, of all the methods tested, it was the one that gave us the best quality DNA when you look at the molecular weight of the DNA.
How has the Siemens platform increased the throughput of FFPE nucleic acid extraction?
On the Siemens system we can do between 48 and 96 samples a day — 48 if you don’t rush it and 96 if you really push it. By hand, technicians can do between maybe 12 and 24 samples a day if they use a faster throughput kit. And the Siemens system requires much less hand-on time than a manual method, and it automates the deparaffinization step — you just put the FFPE curls in a tube on the instrument and start. You have to have some interaction with the robot in between steps, but it's very little compared to a manual method, where you have to use a solvent to de-paraffinize, and then do the extraction by hand.
And how does the quality of the extracted nucleic acids compare to manual methods?
We tested it against a few different methods [from undisclosed vendors] … including … our best manual method. The Siemens system definitely had a higher yield of extraction for DNA and RNA; and the quality of DNA extracted was better than the manual method. The manual method was very good with RNA, but the Siemens system was as good, as well. So there really wasn't any compromise in quality using the Siemens platform compared to our best manual method.
How about cost? What are you pitching to potential customers in terms of cost savings when compared to their manual laboratory methods?
We think there will be cost savings. For one thing, the extraction method is very good. For us, it's the best method we've tested. There might be other methods that we haven't tested that are as good. But the quality and size of the nucleic acids is really beneficial. The de-crosslinking is very good. When we do PCR and real-time PCR amplification tests, the extracted material does really well when you look at how long the fragment is that you can amplify.
In general we do very extensive QC on the extracted material, and some labs might not have the methods, instruments, or expertise to do this.
I can safely say that we're going to offer this for sub-$100 per sample, so it will be significantly below $100 per sample. That might seem like a lot, but if you look at how many hours you have to spend per technician to do manual extraction … even with the cost of consumables being around $10 or $20, you'd still spend a significant amount of time and labor, and then you have to add QC costs. So I think this is a realistic price.
What types of downstream analysis methods do you expect to be extracting nucleic acids for?
Basically, in our center, we have pretty much every technology that is significant in the genomics space. We do microarrays, next-gen sequencing, capillary sequencing, real-time PCR; we have a NanoString system. Any applications in sequencing, including whole-exome and whole-genome sequencing. We're planning to do clinical sequencing, cancer panels and other panels. And certainly [for] a lot of these genomic methods, the input could come from FFPEs. For example, if you look at biomarker studies or clinical trials, certainly you can imagine doing whole-exome or whole-genome sequencing, for example, for looking at risk factors for developing cancer. Using samples from patients before and after treatment you could do follow-up studies. We're actually developing these methods for FFPEs, as well. We already did whole-exome sequencing of FFPEs successfully, and now we're developing whole-genome sequencing of FFPEs. We're trying to optimize our workflow so that we leverage the quality of the DNA we get from the Siemens system in downstream applications. And we're looking at low-input methods; for instance, we may be able to process laser-capture microdissected materials. We're working on this … and it's a bit of a challenge because of the very low amount of nucleic acids you have to work with.
Is there any difference in the sample prep protocol or the method used to extract nucleic acids from FFPE samples depending on the downstream analysis method?
For any method that you're using, you try to get the best quality nucleic acid. It makes more sense not to compromise in the extraction. We have one system where you could potentially use crude extracts from FFPEs – the NanoString nCounter system. We haven't really tested that, and I still feel personally that it's better if you have some quality control measures before you go into a downstream method. So I prefer to extract the NAs, QC them, and then put them on the NanoString system.
For NGS, once you have the DNA, you have to shear it, and then depending on the grade of fragmentation of the DNA you have to adjust your shearing conditions. If the DNA is already fragmented, then you would have to fragment less than you would with fully intact genomic DNA, for example. We're optimizing that right now. From our experience the adaptor ligation and amplification steps of the NGS sample prep are somewhat sensitive to the changes that formalin fixation causes on DNA. Hence the overall efficiency of the NGS sample prep might benefit from a good FFPE DNA extraction method.
In general, you look for the best quality [nucleic acid] that you can extract, so you don't have to compromise on your downstream applications. If you already have issues with your extraction, and then you use a downstream method that is sensitive to contaminants or nucleic acid lesions, you compound the problem.
Will you have the service up and running in a week or so?
I would say within a couple of weeks, we're hoping to be ready.
Do you have any early potential customers?
We have some in Florida. Siemens also, when they talk to customers about the platform, not everybody is interested right away in buying a robot, but they want to see how it performs. And one way to see how it performs is to look at the extracted nucleic acid, so we might get some of that business, where we basically extract some samples for customers who are interested in buying the system.
One other important thing to mention is that we were looking at other providers [for] FFPE extraction services. There are some … but mostly they offer the extraction with some downstream application — for example a diagnostic test. We couldn't really find any place where they only perform FFPE extraction then send you the material to do whatever you want with it. I think that's one of the areas we stand out. We just send you the extracted material and you are not linked to a downstream application.