Skip to main content
Premium Trial:

Request an Annual Quote

Q&A: Jamie Platt of Quest Dx's Nichols Institute on Sample Prep in Clinical Sequencing


Jamie_Platt_QA.jpgNAME: Jamie Platt

TITLE: Scientific director, advanced sequencing, Quest Diagnostics Nichols Institute

At Cambridge Healthtech Institute's Sample Prep & Target Enrichment in Molecular Diagnostics conference last month in Boston, clinical laboratory representatives and technology vendors convened to discuss the current state of sample prep in molecular diagnostics and what issues still need to be addressed.

As chair of the conference session on nucleic acid extraction and sequencing, Jamie Platt of the Quest Diagnostics Nichols Institute provided an overview of the differences in sample prep consideration for research and clinical sequencing studies, and moderated a discussion on the importance of validating nucleic acid sample prep in clinical testing.

Platt noted at the conference that there was a "huge unmet need" when it comes to sample prep in molecular diagnostics, particularly for next-generation sequencing-based assays.

PCR Insider caught up with Platt after the meeting to further discuss her experiences with sample prep for clinical sequencing and how her laboratory at Quest's Nichols Institute is tackling some of the major issues plaguing the field. Following is an edited transcript of that interview.

Tell me a little bit more about the Quest Diagnostics Nichols Institute and your role there.

Quest Diagnostics actually has two Nichols Institutes — one in Chantilly, [Va.], and one here in [San Juan Capistrano, Calif.], where I'm located. The one in San Juan Capistrano was started as a research and development lab to ensure that physicians had esoteric tests available to them to better patient healthcare. Through … acquisition it has become part of Quest Diagnostics, and [Quest] subsequently acquired the Chantilly facility, as well.

We're still focused primarily on developing the testing menu for Quest Diagnostics. We do a lot of laboratory-developed tests, especially for things that are considered esoteric, or highly technical, or specialized.

As scientific director of advanced sequencing, I oversee the technical aspects of both our tests that are … offered on our menu, as well as the tests we are developing and validating. Specifically, I am focused on tests that use next-generation sequencing, [although] we actually use the term 'advanced sequencing,' because it covers all the iterations of sequencing that have followed [NGS].

Your lab is involved in both the development and administration of sequencing-based clinical assays for Quest?

Correct. Anything that is done by advanced sequencing, I have the technical oversight for their administration, and we have specific medical directors that are responsible for overseeing those tests from the clinical side and providing clinical interpretation.

That makes sense in the context of your presentation at the Sample Prep conference, because you're at this crossroads of moving a test from a research-based assay to something that needs to be implemented in the clinic.

Yes, and that's one big challenge we have here. There are lots of different types of clinical labs, from smaller, boutique laboratories that are very specialized, to, in the case of Quest Diagnostics and Nichols Institute, very high-volume reference labs. And we also have a different payor model than some of the smaller laboratories that may take private payors. A lot of our work is based on Medicare, Medicaid, and insurance payors. Except through [Quest division] Athena Diagnostics, we don't offer privately payed clinical reports.

This is probably true across all types of clinical labs: At the conference you said that there was a 'huge unmet need' when it comes to sample prep. I'm assuming you meant for advanced sequencing-based tests, but perhaps you also meant molecular diagnostics, in general. Can you expand on this?

I feel that the unmet need is: Do we recover the target in sample prep? What's happened now, even with digital PCR and advanced sequencing, is that we have a lot of technologies that are very sensitive. Clinicians are expecting us to get results from very little sample, or to get a result that would require great sensitivity. But if you don't prepare the sample correctly, and the target is not in your sample, no matter what you do with downstream technology you're not going to be able to get at the answer.

In terms of sample prep, I feel that's where the huge unmet need is — being able to extract efficiently from clinical samples such as [formalin-fixed paraffin-embedded tissue] or to extract larger volumes from samples — for instance, if you're looking for circulating cells, either circulating fetal cells or tumor cells. There is still an unmet need in being able to deal with the breadth of clinical specimens and also the amount of target within each of those specimen types.

Regarding that, advanced sequencing has the potential to be an extremely sensitive clinical assay, but you said that sensitivity can be compromised or even non-existent if sample prep isn't considered. Do you have any specific experiences related to this?

Yes, we do occasionally get requests from clients … We do a lot of clinical trials business, as well. And one of the requests we often get from our pharma clients is to build tests. Especially in the areas of infectious disease and oncology, one of the requests we get is that they would like greater sensitivity from a limited amount of sample. We often do see requests to use NGS to get a more sensitive result based on a low sample volume … or a limited amount of specimen. For instance, in the case of HIV, you might have 500 microliters of sample at a viral load of 200 copies per milliliter but want to detect mutant virus present at 1 percent. This means that when you extract the 500 microliters of specimen you are theoretically left with 100 copies of virus, given perfect extraction efficiency. If you now want to detect the 1 percent of the population that's mutant, that means you are looking for a single copy. In theory, you would then need to use the entire extract in PCR, with perfect amplification efficiency, to recover that single mutant copy.

That's probably the most specific example I can give, and we've had that with multiple assays and test requests. This example can be extrapolated to other clinical areas involving interrogation of populations, including cancer, RNA-seq, and non-invasive prenatal testing.

We've even had this issue with sequencing tests for influenza, where the specimen type is not ideal to begin with, and typically you don't have much target there. And you get requests to sequence the virus that may be there, but keep in mind there are probably only one or two viral particles in the entire sample. Actually getting a full sequence from the submitted sample is very challenging.

In oncology, the thing to keep in mind is, we really have to understand what the clinical significance is. It's great to say, 'OK, there was one copy there that has this mutation that may be associated with drug resistance.' But what does that mean, clinically? Is that going to actually change the clinical outcome of the patient if you change the therapy based on what you found?

You talked about a lot of the differences that one needs to make when considering sample prep for sequencing research versus clinical assays — the specimen type, nucleic acid extraction method, reagents, equipment, controls, software, et cetera. Are any of these areas in your mind larger gaps than others when translating research assays to clinical assays?

If we really look at the difference between research and clinical [assays], I would have to say the largest gap we have right now is in terms of the software and bioinformatics pipeline. This is because in a research setting you can go through all the literature and hand pick which mutations you feel are important. But in the clinical setting you really have to validate that entire pipeline — the analysis through the interpretation and reporting — and figure out what the most significant call is.

And a different point on that: Often times people are using terms like 'clinically actionable' to determine if mutations should be reported or not, but that's even a tricky definition, because in some cases the clinical action may be to do nothing. So I think that's a bit of a misnomer. I've heard that argued in several different sequencing discussions with key opinion leaders — how do you define clinically actionable. For instance, if a patient is terminal and there is no known treatment, clinically actionable may be just enjoying the rest of the days you have without unnecessary treatments that are going to affect quality of life.

You also discussed some issue with nucleic acid extraction bias, noting that methods with the highest yield and purity may not result in the recovery of all targets; and vice-versa, that the methods resulting in the recovery of all targets may not have the highest yield and purity. Can you explain that a bit more?

Yes, several research papers have looked at nucleic acid purity and nucleic acid extraction. I think, initially, it seemed like people generally thought that if you had good yield and purity from a nucleic acid extraction method that you'd be successful in recovering your target. However, as some papers have pointed out — and one of the better examples is the Guo and Zhang study [Appl Microbiol Biotechnol. 2012 Jul 4. Epub ahead of print] — that even though the purity and yield may be good on face value, for some reason certain targets aren't recovered as effectively.

I think you have to specifically look for the target you're trying to recover in order to really assess nucleic acid extraction success. You have to use more than just yield and purity to understand whether it's the best method for your test.

Do you think that tech vendors are doing their part to address sample prep for clinical sequencing? What would you still like to see from vendors in this area?

In general, some vendors have done a better job than others. I think one of the biggest needs for a clinical lab, especially a large-volume clinical lab, is automation of sample prep. In terms of bringing these technologies to the clinic, things need to be automated as much as possible from A to Z — from sample prep through the bioinformatics and even reporting the key mutations.

Obviously work is being done in that area. And when it comes to laboratory-developed tests, it's probably a little more challenging for the vendors to address that, because a lot of those labs want their own flavors and spins on things. But I think in general it's moving in the right direction, and some vendors are moving forward with [US Food and Drug Administration] submissions that encompass everything from sample prep through sequencing and reporting. But it is definitely one of the most challenging points.

Another gap is in reagents and improvements. In any technical area that is advancing really rapidly, whether it's digital PCR or advanced sequencing, a challenge is that vendors are constantly improving their chemistries and detectors, and they want to give people the latest and greatest. The research labs can move ahead with things like that, and take advantage of those improvements, but in a clinical lab we have to have things locked down. So version changes and new reagents and software and detection systems are really difficult for the clinical lab to deal with. That's a pain point for clinical labs: the nature of rapidly evolving technology and judging when to jump on board with that technology and having to do multiple validations; or determining to wait a little bit longer until the vendors get some of those details ironed out.

Are there any specific sample prep or automation technologies that your lab has employed or evaluated recently, or is aware of coming down the pipeline, that could greatly improve your workflow?

It's very early, but one of the things I'm most excited about is sequencing technology that essentially won't require much sample prep or will all be automated. There are several emerging technologies where you'll actually be able to put the sample on, and then just get the answer out on the back end with very little handling of specimen in between. There are several companies working on that. I think they're still a few years out, but those types of things are pretty promising.

Even though your lab is heavily focused on advanced sequencing technologies, you mentioned digital PCR several times in your talk and in this interview. Is the clinical potential of that technology rivaling sequencing? Do you expect it will be a 'one or the other' situation, or will they complement each other?

I think a little bit of both. Depending on what you're trying to look at, you can do a digital PCR assay rather than an advanced sequencing assay, for instance if you're looking for certain targets or SNPs. It's definitely a little more straightforward in terms of implementing in a CLIA lab setting.

The advantage of sequencing, and even Sanger sequencing, is you always get more information. In my opinion you can build much better quality around sequencing assays, because there are so many more data points in the end. For instance, for things like our HIV genotyping, we have built huge databases. We essentially do a phylogenetic tree and then use a BLAST to look at previous samples we've run. And we've been able to build a good QC metric into our samples, because we can see if there is any cross-contamination or carry over from batch to batch or sample to sample.

If you've built the right QC into your test, you have better control of quality with sequencing that is just not possible with any PCR assay, even digital PCR.

What's the overarching takeaway from your talk on sample prep in clinical sequencing?

You can’t get blood from a turnip, and I've seen some reports and papers from other laboratories where they really didn't control the whole process sufficiently and don't have a good idea of what their overall assay sensitivity is. They may claim sensitivity based on the detection platform, but not on the entire assay. When you report a clinical assay's sensitivity or limit of detection, it has to consider the entire process from the time you get the specimen to the time it comes off the detection platform and gets analyzed. You can't just say, 'NGS can detect down to 0.1 percent, so our assay is 0.1 percent sensitive.'