Serum miRNA Biomarkers for Early Detection of Ovarian Cancer
Prof. Dipanjan Chowdhury is Associate Professor at the Dana Farber Cancer Institute, part of the Harvard Medical School in Boston, United States. He is investigating the role of miRNAs in DNA repair. Over the years, his lab has used Exiqon's miRNA NGS and qPCR Services (now QIAGEN Genomic Services) in several different projects.
What is the main focus of the research conducted in your lab?
The crux of my research interest is genome stability and DNA repair. DNA repair is relevant for both cancer therapy and the process of oncogenesis.
How did you come to be interested in miRNAs?
I started my lab back in 2007 when miRNAs were a relatively new class of molecules, and at that time no one was looking at the role of miRNAs in DNA repair. So we wanted to understand the role of miRNAs in the relatively well-studied process of DNA repair.
How has the approach to miRNA research changed over the years?
Back in 2007, understanding of miRNA functions was still in its infancy. We started with lab-based experiments: finding a miRNA, and then asking if it is clinically relevant. Nowadays we do things the other way around, we first start with the patient samples, we identify the clinically relevant miRNAs, and then investigate the mechanism by which they impact the process we are studying.
Now with so much greater availability of clinical samples, and new technologies enabling sequencing from very small amounts of material, we can take the research to a new level.
We started off studying tumor-based miRNAs, and indeed we still do, but the serum based miRNAs are going to be much more powerful as biomarkers, as it’s ultimately a blood-based non-invasive test.
Of all your projects involving miRNA, which do you find most exciting and why?
We are doing a lot in the realm of serum miRNAs as biomarkers. I think the most exciting study in my lab using Exiqon Biofluids miRNA NGS and qPCR Services (now QIAGEN Genomic Services) is to find out if we can use serum miRNAs for early detection of ovarian cancer. Ovarian cancer is the most lethal gynecological disease in the United States. Early detection of ovarian cancer is crucial for the best patient outcomes. Unfortunately there is no early detection test for ovarian cancer, and the symptoms of ovarian cancer are very weak. So we set out to identify miRNAs in serum indicative of ovarian cancer, with a view to developing a blood based test for ovarian cancer.
What were the results of your project to identify serum biomarkers for ovarian cancer using our miRNA NGS and qPCR Services?
We sent 180 patient serum samples for small RNA sequencing. The patients had ovarian cancer in different stages, benign tumors or no tumor. It's really exciting because we have identified a set of miRNAs that are very promising: the serum miRNA signature can distinguish benign vs. malignant disease – and can even distinguish stage I cancer vs. benign disease. This is a big deal because when the patient comes to the clinic there is currently no way of telling whether or not the patient has ovarian cancer, until they do the surgery and the biopsy, which is an extremely invasive procedure. Using the miRNA biomarkers we can also distinguish early disease from late disease.
Could you tell us about your projects on serum miRNA biomarkers to assess the impact of total body irradiation?
We published a paper last year (2) using mouse models where we used your miRNA qPCR Services to identify two serum miRNAs that can help us predict the outcome of total body irradiation. In today’s world, with the current political climate and terrorist threat, we have to plan how to respond to a nuclear disaster. Right now there is no test to identify whether or not a person has been exposed to radiation. Based on the miRNA biomarkers in serum, we can actually identify which animals were exposed to the radiation, and we can predict the impact of radiation exposure, i.e., whether or not the exposure will be lethal.
We have recently done a similar study with your miRNA qPCR Services using existing blood samples from macaques – these animals are individuals (not inbred), so they are the closest you can get to the real human situation. This study is actually very powerful because the miRNA biomarkers we identified in mouse are evolutionarily conserved and did hold true in macaque. These miRNA biomarkers could be used to identify individuals most likely to benefit from drugs to mitigate the effect of radiation.
Did the macaque study on serum miRNAs reveal any new information?
What’s interesting about the macaque study is that the same dose of irradiation is lethal for some animals but not for others. This highlights that individuals respond differently to the same dose of irradiation, and we could predict the response to irradiation using just two miRNA biomarkers. It's incredible!
There also appears to be gender differences as well – females seem to be more radiation sensitive – which is a new observation. There is currently no systematic study on the gender aspect of radiation sensitivity because animal studies usually include only one gender to minimize variability. We were lucky that this macaque cohort included both genders. One miRNA basically showed us the gender bias. These are really interesting observations and also raise so many questions: where are these serum miRNAs coming from? What are they doing? Where are they going?
Can serum miRNAs also be useful biomarkers for radiation therapy?
We are looking at radiation therapy and whether serum miRNAs can predict the later development or type of secondary tumors. We have a mouse model where mice are irradiated at a certain dose and go on to develop lymphomas or sarcomas 4–5 months later. We have identified two miRNAs, present in serum one week after radiation, that predict whether the secondary tumor will be a lymphoma or sarcoma. That data provided us with the basis for a project we are currently working on, using human patient samples.
What were some specific challenges in your projects?
One of the biggest issues when developing any biomarker is the sample numbers, and having access to clinical samples with all the relevant clinical data available. You have to look at hundreds of samples from different cohorts to really make sure you are on the right track.
Another challenge is sample amount. It’s very hard to get hold of these clinical samples and sometimes the amount we receive is very limited. The QC checks that you do for biofluid samples are good to do, but I would like it if the QC could be done using less material.
Another big issue with serum miRNAs is normalizers. It’s not simple – there’s no GAPDH, there’s no actin. We’ve spent a lot of time figuring out what is the best normalizer. To be honest, it’s not as simple as just run NormFinder and identify the best normalizers.
How did you overcome the challenge of normalization in serum?
You really need to run these algorithms, identify 10 or 12 candidate normalizer miRNAs, and then actually test them to see which do not change. In the mouse serum study we spent a long time doing it, and the same with the macaque and the human projects.
What has been your experience validating miRNA NGS results by qPCR? Did you find a good agreement between the two platforms?
14 serum miRNAs identified as ovarian cancer biomarkers by your Services team by NGS were subsequently validated by miRNA qPCR. The first validation project was done using 120 serum samples (the same samples that were used for sequencing).
Overall we found good agreement between the two platforms, however we did identify some miRNAs by NGS which were not identified by qPCR. This could be due to different thresholds applied during the NGS and qPCR analysis, or different sensitivity levels of the platforms.
Nevertheless, the miRNAs that were successfully validated by qPCR still constituted a statistically significant signature for ovarian cancer, so the qPCR validation essentially refined the signature. Ultimately, any diagnostic test will be based on qPCR and not NGS.
We have also sent serum samples from an independent cohort of 226 patients for analysis by qPCR, and those results have literally arrived today so it’s exciting.
Why did you choose to use our Services?
We have worked with Exiqon for many years. This has been very critical; I think the miRNA qPCR platform is very good. I’ve tried many platforms, trust me! I am totally neutral, but I do think the LNA qPCR platform is very good – particularly with regard to specificity, which is crucial.
I believe you have conducted some of the experiments in your own lab, and some with our Services team. What factors do you consider when deciding whether to send samples to us for analysis, or perform the analysis in your own lab?
Honestly, we send all the human samples to your Services team, and any precious samples with limited amount of material available we always send as well. We feel that the Services team are the professionals so we prefer them to handle those samples. In our lab we mainly handle tissue culture based samples.
My lab is relatively small by choice, so I prefer that the Services team takes care of the samples they know how to handle best – and it’s easier for me to send samples to them than have one person spend a month doing it in our lab. We would rather spend our time doing the functional analysis.
The ovarian cancer project, the macaque project – these projects have literally not been touched by anyone in my lab; we just got the samples and shipped them to the Services team, and I really like that, honestly, because we send the serum and they do everything everything. If it's good, it’s good, and if it’s not good it's their fault, so I can blame them (which hasn’t happened so far, so it's all good!).
What are the next steps in the ovarian cancer project?
Now we are trying to get more samples to solidify the results we have for early detection of ovarian cancer in serum. We would also like to find out more about the 14 miRNAs we have identified as biomarkers of ovarian cancer: do they have a role, what are they doing? Once the clinical application has been demonstrated in further cohorts, we can begin to go after the biological questions and really find out what these miRNAs are doing. It’s exciting times.
What are the future perspectives for the ovarian cancer project?
It would be very exciting to analyze serum samples from patients before they develop cancer. There are some very precious collections of blood samples taken from thousands of individuals, and a small number of them ended up getting cancer. If we can sequence these samples, maybe we can develop a blood based miRNA test that predicts ovarian cancer before onset of the disease – that's the ultimate diagnostic. This will be one of the most important things – if we can do it, it will be a huge breakthrough. We are cautiously optimistic because we definitely have a test that distinguished benign vs malignant disease. I genuinely believe this is going to make a difference.
In your opinion, how large is the potential for miRNA biomarkers?
There's certainly a huge potential for miRNA biomarkers. I also believe a lot of the potential has been untapped because there’s a lot of mis-information and wrong data out there. Many of the problems have to do with platforms, and not taking care with the analysis – like the issue of normalizers for serum miRNAs for example. This diffuses the enthusiasm for the field for outsiders or clinicians who want to use it.
So it’s really very important for us to do it right, repeat it, and really convince people that miRNA biomarkers have the potential. I really believe that the blood based miRNAs (plasma or serum) have a huge potential as biomarkers. And for many cancers they have not really been looked at in a very rigorous way, so there is definitely a huge potential in this area.
What are the advantages of miRNAs, from a biomarker perspective?
I think miRNAs are great biomarkers – there's not many things that are that stable, so small, you can detect them easily – PCR based detection is so much easier than finding antibodies. With one PCR you can detect any miRNA. So you just need to analyze different miRNAs, but using the same overall tools and technique.
How do you see these ovarian cancer miRNA biomarkers being developed into a test for the benefit of patients?
The big picture here is to come up with a blood test that can be part of a woman's annual physical check-up. So as you go and get your cholesterol and sugar levels checked, you can have the miRNA levels checked too, to see if you are at risk of developing ovarian cancer. We’ve developed an algorithm where you punch in the levels of each of the 14 miRNAs, and this gives you a score for the probability of having ovarian cancer.
There is certainly a lot of interest from clinicians here. If this pans out, the application for this test will be very broad. I will be very happy to see these miRNAs put to use in the clinic for the benefit of patients.
When and where will we read more about your studies?
We have two papers currently under review – one on the miRNA signature for early detection of ovarian cancer in human serum, and the other on the macaque study looking at serum miRNA biomarkers to predict the impact of total body irradiation.