Muneesh Tewari, a former Fred Hutchinson Cancer Research Center scientist who recently moved to the University of Michigan, has taken a special interest in microRNAs and their potential as biomarkers for disease.
Earlier this month at Cambridge Healthtech Institute's Molecular Medicine Tri-Conference in San Francisco, Tewari presented some of his recent work in this area. He highlighted the fact that research on circulating miRNAs as cancer biomarkers is now nearing the six-year mark, and reviewed what researchers have learned and where they go from here.
PCR Insider caught up with Tewari after the conference to discuss the challenges and hopes for miRNA and droplet digital PCR as an enabling technology, among other topics. Following is an edited version of the conversation.
You recently moved from the Hutch to University of Michigan. What is your new title, and how's the transition going?
I'm an associate professor in the department of internal of medicine, specifically in the division of hematology/oncology. I also have a joint appointment as an associate professor in the department of biomedical engineering. I'm a member of [U of M's] Biointerfaces Institute, which is sort of an engineering, biology, [and] medicine collaborative group. I've also got a membership in the center for computational medicine and bioinformatics. So, it's a very interdisciplinary position and opportunity. I went to medical school and graduate school here, and I also did residency here in internal medicine. I left Ann Arbor about 15 years ago, so I have a history here.
How did you get started studying miRNAs. What do you like best about them?
I got started studying miRNAs because I came from a systems biology background. I had been studying protein interaction networks. What really attracted me about miRNAs [was], whereas a lot of protein-protein interactions are not necessarily regulatory, but sometimes help make complexes, or molecular machines … inherently the miRNA-target interaction is regulatory, or at least it is in the many examples that we've seen in biology so far. This is really a regulatory interaction between two biomolecules. That's what got me interested initially, because I figured an interaction that is regulatory and molecule that's regulatory is likely to be important. At the time I was thinking it would be important as a biomarker, and potentially in terms of thinking about therapy.
Since then, I think what has been especially interesting, from a more translational and biomarker perspective, is their pretty unusual stability compared to lots of other nucleic acid molecules. That said, I try not to be miRNA-centric. My real interest ultimately has grown into disease detection and disease monitoring, and even ultimately, hopefully, into health assessment and health monitoring. I think for a lot of the big problems that we face in healthcare today, it's going to require a very holistic approach at a molecular biology level, and I think miRNAs may be a very important part of that bigger picture.
I'm wondering about the state of the field of miRNA analysis. In your Nature Reviews Genetics paper [from May 2012] regarding profiling miRNA, you said there were about 1,000 mature miRNAs in humans "according to miRBase accessed January 22, 2012." A similar search today says there are "1,872 precursors, 2,578 mature" ─ how many of these do you think will prove to be functional, and did you expect the number to grow so substantially back in 2012?
I'm not surprised that the number has grown. With the application of next-generation sequencing to various types of biological samples, many more small RNAs have been identified. At the same time, it's an open question how many of those are actually functional, since most of these new additions tend to be pretty low-abundance RNA. As you sequence more deeply, you tend to find more small RNAs. Some could be miRNA in the process of being tapped by evolution. That's one of the ideas out there ─ that we might be finding miRNAs that may be recently born but have yet to become expressible at higher levels and adopt a function.
It is a pretty ancient form of cellular regulation. The first miRNA was discovered in [Caenorhabditis] elegans, and there are miRNAs that function similarly in plants. It's an ancient form of regulation but currently, of the number of putative miRNAs that have been identified, how many of those are functional, relevant, [or] active, and how many are simply small RNAs that have been transcribed but may not have a function? Most of the new ones that are found and continuously added are not conserved throughout evolution. I think most of the ancient, evolutionarily conserved miRNAs were found early on. miRBase is a community repository of miRNA sequences. At the same time, all of the sequences that are there, although they may meet some criteria of being miRNAs, not all of them have equivalent evidence for being functional. A lot have a lot less evidence for being functional that some of the original ones discovered.
In this review, you listed applications of miRNA profiling. You wrote that list about a year and a half ago ─ have you been surprised by the pace of the field since then? Has any subset of applications (novel miRNA discovery or circulating biomarkers, for example) progressed in particularly interesting or unexpected ways, or have there been any new applications that you did not anticipate?
I'm not aware of any especially new application. I've just been surprised at the pace, the number of different applications, the amount of research going on, and number of papers being published with respect to circulating or other extracellular miRNAs in diverse areas, including forensics, as well all kinds of disease states, and even pretty rare diseases.
About detection technologies, in that article you listed various qRT-PCR platforms or assays, but this table didn't include ddPCR. Last October, you published a Nature Methods paper on ddPCR. Can you tell our readers more about how ddPCR is used for miRNA analysis, your professional opinion about this technology, and how it has changed the way you work in the lab? You also recently published in the Journal of the American Society for Mass Spectrometry on using high-resolution mass spec to analyze miRNA in samples ─ how does this compare to PCR-based platforms, and are those two competing technologies in your mind?
They're not competing at all. I think they're in different stages of their development. Whereas the mass spec is very early stage, proof of concept, methods development, I think droplet digital PCR is much closer to real-world application. It is already there in the research setting and moving in the direction of real-world application in the clinical setting.
In terms of how ddPCR can be used for miRNAs, it's basically a technology for absolute quantification of miRNAs that doesn't require standard curves and that is highly reproducible day to day when compared to what has been historically the gold standard, which is real-time quantitative PCR. In my lab, how has it changed the work? It hasn't replaced all the real-time PCR that we do, nor should it. But it has removed a major roadblock to taking the next step toward multi-institutional and prospective circulating miRNA biomarker trials, to measure these in patients day to day as they come in. This was a major roadblock in my mind for moving this field forward. I think because of the reproducibility and the pretty high precision at low concentrations of the target, ddPCR has the potential to make it possible to do the next wave of clinical trials with circulating miRNAs as biomarkers.
For what applications in the lab do you prefer the 'old fashioned' way?
For applications where absolute quantification is not important at all [or] for cases where all you need is relative quantification — comparing two samples to get a ratio of how much is present in one versus the other — one might prefer real-time PCR currently, just because the workflow is a little bit more straightforward and the throughput is a little bit higher. That's really the main place. And for applications where the precision isn't that critical, one might prefer real-time PCR. Again, that's based on the current iteration of ddPCR. In the future, if the throughput and workflow for ddPCR becomes similar, or identical, or maybe even better than that for real-time PCR, then this might change.
In your Nature review you also mentioned frontiers in technology and methodology, including whether enrichment is ultimately necessary and methods for quality control or normalization. Has there been any development here?
In general in my lab we don't enrich for small RNAs before doing miRNA quantification by real-time PCR or ddPCR. We have done that before doing next-generation sequencing, and it's been quite important and valuable there. Size fractionation and gel purification, in the research setting, can be pretty useful. But it is labor intensive and it introduces sample-to-sample variability. That's the reason we don't use it for a lot of our clinical studies where we're trying to really minimize the variability.
That said, in principle, size fractionation should be beneficial and if there were better technologies or approaches available that were higher throughput and also very reproducible, it would be worth considering. I think that's an open area that's worth developing.
In terms of assessing extraction efficiency, I'm not aware of anything better than a spike-in control. For quality and quantity, if you have enough miRNA, there are Agilent Bioanalyzer chips that can look at that. Frequently, when looking at miRNA from plasma or serum, from patients and from relatively small volumes, it's not enough to be analyzed that way. I think assessing quality and quantity is actually kind of a controversial area because it also gets into normalization, how to normalize for differences in quality and quantity. There have been many approaches proposed, but at this point it's still not completely clear. I think the spike-in controls make sense, but I think this is still an open area for more investigation.
A lot of researchers have been excited as banks of formalin-fixed paraffin-embedded tissue samples suddenly become accessible to more molecular biology techniques. Have you seen increasing numbers of papers analyzing miRNAs in FFPE? You'd also mentioned in your review that plasma was a problematic sample. Do you know of any new products or techniques for this?
FFPE is a very enticing and potentially great type of sample when thinking about miRNAs, because of their stability. I have seen a trend towards more FFPE studies, but I've been surprised that it hasn't been even greater because of the opportunity that is actually there. It might just be that it takes some time to catch on.
In terms of new products for assaying miRNAs in plasma, there are a number of commercial products. I couldn't say if there is anything revolutionary that I'm aware of right now. In our lab we've used various kits over the years, including the Ambion [Life Technologies] mirVana kit and the Qiagen miRNAeasy kit. More recently we've been using the Exiqon biofluids kit. They each have strengths and weaknesses. It's still a somewhat intensive process and we still have to worry to some extent about the challenges of getting highly pure RNA.
In your ddPCR paper, some of your co-authors were from Bio-Rad, so I'm going to guess that's the platform you use in the lab. Why this platform? Are you aware of any studies comparing different ddPCR platforms head-to-head?
For ddPCR we do use the Bio-Rad platform. I don't have any strong opinion on others, because I just haven't had personal experience with them. I may be investigating them as time goes by. There aren't any head-to-head studies that I'm aware of, because ddPCR as a commercially available technology is pretty new. The Bio-Rad platform, as far as I'm aware, was the first one. My lab was involved with that as beta testers from the company QuantaLife, which was then bought by Bio-Rad. So we really started this as a collaboration, testing this new platform that was being developed by QuantaLife.
As far as I'm aware, RainDance [Technologies] subsequently came out with a commercially available version of their droplet platform. They've been into droplets for some time, but developed their droplet digital PCR platform subsequently. I'm not personally aware of any direct comparisons that have been done between the Bio-Rad and RainDance platform. Certainly not for miRNA ddPCR, but I'm not even aware that they have been compared for any ddPCR.
People like to use the term disruptive technology; has ddPCR been a real breakthrough in your lab, and might it qualify as disruptive?
In my lab, yes, ddPCR is actually giving us very clear answers where in the past we were getting fuzzy answers. Has it been a breakthrough? Yes. Within that area of being able to perform absolute quantification of miRNAs in samples where the concentration is low, it has been a breakthrough. With respect to being a disruptive technology, disruptive on what scale? I think it's a bit early to say right now whether it's disruptive on a large scale. I think it certainly has the potential to be disruptive in the circulating miRNA diagnostics field. I think it certainly has that potential. Over time, it'll become clear.
You recently published a study in PLOS One of serum miRNA biomarkers for prostate cancer. Can you summarize the results and tell our readers where you're going next?
In that paper we did discovery of miRNAs in the serum of advanced prostate cancer patients to identify miRNAs that might be associated with advanced, metastatic, prostate cancer. We found roughly half a dozen miRNAs that were elevated in a substantial fraction of patients with advanced prostate cancer compared to age matched men who didn't have prostate cancer, at least to the extent that we could evaluate.
What I think was most interesting is, when we looked at those miRNAs, one of those was a miRNA called mir-210, which is a hypoxia-inducible miRNA. It made us wonder whether the levels of mir-210 in the serum of these men might reflect the level of hypoxia that the tumor tissue might be experiencing. Although we couldn't definitively determine whether the mir-210 was coming exclusively or largely from the tumor tissue, we were able to ask if there was a relationship between the levels of serum mir-210 and whether the disease in those patients was responding to treatment or not. This was retrospective analysis, so it requires future studies to validate or confirm the findings, but basically we found that patients who had high levels of this hypoxia-induced miRNA in their serum tended to be patients who were not responding well to therapy, regardless of what therapy they were on. This is consistent with prior tissue based studies which suggest hypoxia renders a variety of cancer types generally resistant to therapy, including radiation therapy and a variety of chemotherapies.
With respect to where we go from here, I think the interesting thing is to try to validate this hypothesis that the serum levels of mir-210 reflect the prognosis, or how well a treatment is working, and to determine whether mir-210 levels could actually predict whether a patient is likely to be responsive to a therapy or not. That said, this is still a pretty early proof-of-concept study, but I think it opens up new questions and that's its value.
Mir-210 has evolved as a miRNA that helps cells adapt to low oxygen conditions. Many of its targets need to be repressed as part of the adaptation response to low oxygen. One of the exciting things about that study is also the concept that serum miRNA markers may reflect dynamic changes in tumor physiology. Circulating DNA markers, while I think they can [also] be really important as well, tend to be relatively constant over the course of disease, aside from accumulation of certain mutations. The fact that mir-210 is hypoxia-inducible raises the possibility that some of these miRNAs might be able to reflect what's going on physiologically in a patient today, compared to a week ago or two weeks ago, and we can try to capture some of those dynamic changes and maybe even capture some of the more local micro-environmental changes. That is if, in fact, the serum mir-210 reflects local tissue hypoxia — that's our hypothesis.
What do you think will be the next big frontier in your field?
It's probably three big things for extracellular miRNA biomarkers. One is just increased understanding of how they're turned over, how they get into the circulation, and whether they have any functions beyond biomarkers. The second is moving all of these biomarkers that have been discovered into validation and cross validation across multiple institutions, to figure out what's reliable and what's not. The third big one is how do you integrate this class of markers with many other classes of markers in a thoughtful way, so that you're learning something biological from the biomarker about the disease process. At least it's my bias that the closer a biomarker is to the biology of the disease the more likely it is to be clinically useful.
In terms of panels, what I meant was also looking beyond miRNAs — how are miRNAs combined with protein markers, circulating tumor DNA, or methylated DNA markers, or other extracellular RNA markers that are being studied, or even combined with risk factors or markers we might not think of as molecular, like behavior or lifestyle or exposures? How do we put all this together? I think there's value in that, but it's also very challenging to know how to do that thoughtfully and effectively.
Do you have any messages for investors or companies or other labs in the field who might be interested in this subject?
My main message is we should all always seek fundamental understanding. Even though we are interested in solving an applied problem and improving health, I think we're going to move forward most effectively if we don't give up the imperative to seek deep understanding. For example, one of the challenges a number of people have run into trying to find serum miRNA markers for cancer is that a number of biomarkers have been discovered that seem, at least in some patient populations, to be higher in cases of cancer compared to controls. But it turns out that in many cases, these are in fact markers of blood cells, and they're likely to reflect differences in blood counts in one group of individuals compared to another. Without having an understanding of where miRNAs are coming from and which miRNAs are really specific to blood cells … it's going to be very difficult to move forward productively. That's just one example of where understanding at a very physical level ─ the miRNAs are there, where do they come from, where do they go, how are they regulated? ─ is important. Even though a lot of these are more fundamental biology questions, I think it's still essential to investigate them and keep investigating them, in parallel with and in order to try to solve applied problems.