transcript imaging in animals webinar 9 dec 2011 in... · about technologies related to...
TRANSCRIPT
1
Determining the Impact of New Therapeutic Approaches: Advancing Imaging in Animals
webinar 7 Dec 2011 [0:00:14] Slide 1 Sean Sanders: Hello and a warm welcome to this Science/AAAS audio webinar. My
name is Sean Sanders and I'm the editor for custom publishing at Science.
For you today, we have the third and final webinar in the series
exploring advances in imaging technology from cell to well to animal. Find the previous two webinars at www.sciencemag.org/webinar. In today's discussion, our two panelists will talk about how the use of in vivo imaging particularly fluorescence tomography allows the study of biological pathways in context, noninvasively, and in their natural state. Imaging agents used for fluorescence tomography are now even being translated into the clinic for human diagnostic applications while disease‐specific fluorescence markers are finding their way back into basic research in high content screening applications.
The panel today will discuss the most important considerations
when performing in vivo imaging and describe how in vivo imaging techniques have been applied in their research. They will also be available to answer questions submitted by our live webinar viewers.
Slide 2 It now gives me great pleasure to welcome our speakers for this
discussion. We have Dr. Patrick McConville from Molecular Imaging in Ann Arbor, Michigan and Dr. Matthias Nahrendorf from Harvard Medical School in Boston, Massachusetts. Many thanks to you both for being with us today.
Before we get started, I have some useful information for our
audience. You can resize or hide any of the windows in your viewing console. The widgets at the bottom of the console control what you see. Click on these to see the speaker bios or additional information about technologies related to today's discussion or even to download PDFs of the slides.
2
Each of our speakers will talk briefly about their work. After which we will have a Q&A session during which our guests will address the questions submitted by our live online viewers. So if you're joining us live, start thinking about your questions now and submit them at any time by typing them into the box on the bottom left of your viewing console and clicking the submit button. If you can't see this box, click the red Q&A widget at the bottom of the screen. Please remember to keep your questions short and concise, as this will give them the best chance of being put to our panel.
You can also log in to your Facebook, Twitter, and LinkedIn accounts
during the webinar to post updates or send tweets about the event, just click the relevant widgets at the bottom of the screen. For tweets, you can add the hash tag, #sciencewebinar.
Finally, thank you to PerkinElmer for their sponsorship of today's
webinar. Slide 3 Now, I'd like to introduce our first speaker for today and that's Dr.
Patrick McConville. Dr. McConville is chief operating officer, chief scientific officer, and a co‐founder of Molecular Imaging Incorporated, the only contract research organization dedicated to in vivo preclinical imaging. He has over 10 years of experience in multimodality small animal imaging with a drug discovery focus, and has specialized in a number of disease areas. After postdoctoral work at the National Institutes of Health, Dr. McConville joined with academic and industry collaborators to found the first multimodality imaging contract research organization, MIR, in 2003. MIR grew to support most of the world’s top pharmaceutical companies through in vivo MRI, PET, optical, and CT imaging capabilities. The company Molecular Imaging was launched in May 2011 and is extending the range of imaging modalities and associated biomarkers and disease models available to industry decision makers. Warm welcome, Dr. McConville.
Slide 4 to Slide 5 Thanks very much, Sean. So thanks everyone for attending and I'm
going to present a CRO perspective on the use of FMT, but I'm going to start by looking at the way imaging has evolved over the last few decades to see its prevalent use in discovery and development from a general point of view. Then I'll get into some disease endpoints and some of the most prevalent endpoints in how these are validated and the concept of biomarker and target marker from an imaging
3
point of view, and then begin to look at the concept of probe‐facilitated imaging, which is where the FMT technology really comes into play. I'm going to finish by presenting a few specific examples for us using FMT and FMT probes for inflammation and tumor burden tracking.
Slide 6 So to begin with, imaging, medical imaging has only been around for
a short period of time really. In the '70s, the major modalities were introduced and it's only recently that these modalities have been specialized and adopted and translated for specialized rodent work. So the last 5 to 10 years has seen a transition from the use of medical imaging in a largely anatomical and diagnostic fashion through to today's use, which is very prevalently in imaging function and molecular imaging. So, we've seen a transition from diagnosis into efficacy and a parallel transition of the use of this technology from something that was pioneered within the academic centers, but that now has become very prevalent in the pharmaceutical industry and the CROs. The next 5 to 10 years, we'll see more and more use of imaging for assessing drug safety and toxicology and not only efficacy.
Slide 7 [0:05:29] So, to summarize some of the major changes and concepts, imaging
was modality centric and was a very decentralized thing where single labs would focus on MRI only for example, but today it's become very modality agnostic and applications are driving the use of imaging. It's a centralized concept today. It was a complex technology that required physicists and engineers very heavily. Fortunately, it still does require those types of people, me being one of them, but it is more and more of a black box technology that multidisciplinary teams can utilize. It was a tool for diagnosis as I said but it's not used very heavily for disease progression and therapeutic response. It was expensive, still is expensive unfortunately, but that will continue to be the case as the providers continue to push the boundaries of the technology so we're seeing very rapid advancements and that has expense associated with it obviously. It was not a standard in drug development, but it now is and as I've mentioned, it has evolved to become a very prevalently and heavily used tool in the industry.
Slide 8 to Slide 9 So, here's the discovery and development continuum for drug
research and the first key point here is that imaging is today used
4
right across this spectrum. So it's a translatable technology and that's something that makes it very important.
Slide 10 But the next key concept here is that the foundation for the use of
imaging today more than ever is really the therapeutic areas, and so I consider these therapeutic areas and here are five of the major ones as the foundation for the use of imaging. These are the towers across which we would like to apply imaging broadly.
Slide 11 But driving the use of imaging really is the pharmacology, the
questions coming from pharmacology teams, toxicology teams, and pathologists, and this is why the applications have become so well validated and used.
Slide 12 So why just generally would we use imaging? We have to make sure
that we are gaining an advantage from it. We don’t want to be using it just because it's accessible. It might be faster enabling early prediction. It could be better resolved at the tissue or sub‐tissue level enabling us to get down to those levels. It might be more relevant by providing access to unique endpoints that aren't accessible through other means, and the translational capability is important. If it is translational and we hope it will be a lot of time then the continuum for discovery and development is addressed and that becomes very powerful. A question often asked is it cost effective. So I'm mentioned before that it is expensive to invest in the technology, but overall its use is cost effective if at least one or more of the above is true in a drug program.
Slide 13 So endpoints and how do we validate image based endpoints. Slide 14 Firstly, we can group some of the major endpoints into categories,
the first being anatomical where imaging is used to diagnose and track disease evolution or progression. We can then get into functional endpoints. Some of them are shown here, true imaging where it becomes more and more powerful and then down at the molecular level performing true molecular imaging looking at pathways and target modulation. So imaging, preclinical and clinical imaging provides this today and this is really a subset of what's
5
accessible in terms of image‐based endpoints. You can see that it's not huge stretch to think about the use of imaging in safety and toxicology even though it's predominantly used for efficacy and diagnosis today. Some of the same endpoints would have applications in safety and toxicology so that's driving that evolution.
Slide 15 So how do we perform an assessment through an image? What we
have at the earliest stage is an image‐based quantifiable parameter. We have an image and we can quantify contrast or signal level in some manner, what we really need to do though is invalidate that. So we provide or perform some level of validation to correlate that parameter with disease progression and that's what I would call validation phase I. If that's successful, we end up with an imaging based biomarker. At that point, we have something valuable that's quantitative that can be used for decision making in drug research.
What we would ultimately love to be able to do is to perform
additional validation so that we might end up with what is known as a surrogate marker. So to do that, we're correlating with a true clinical endpoint and we are ending up with, as I've said, a surrogate marker. So that's really the Holy Grail for any biomarker and more and more into the future, we hope and we believe we will see image‐based surrogate markers.
Slide 16 [0:10:18] So, as I mentioned before, what we want to be able to do is to apply
modalities across therapeutic areas very broadly. So this is the challenge of a core facility and this is the concept I was making before having multiple modalities housed with a single operation and then being able to apply these very broadly. So it's a very powerful concept, but it's a challenging one when you consider the validation that's needed.
Slide 17 So, we would envision having multiple protocols per imaging
modality that creates an additional level of complexity. The multiple therapeutic areas, the towers that I mentioned applying it broadly across all of those, multiple models per therapeutic area, even more complexity and then we get into safety and medical devices. That's ultimately what we'd like to be able to do. It becomes an unwieldy challenge in some ways, but this is the concept.
6
Slide 18 So now, I'd like to get into probe‐facilitated imaging and FMT
specifically and how we're using that in the industry today. Slide 19 So firstly, on imaging probes, an imaging probe can be generally
defined as a molecule or a nanoparticle that's designed to modulate imaging contrast in an image. The degree of that modulation is generally dependent on probe concentration, the voxel level. It's dependent on the access to tissue and the kinetics of the distribution of that probe and its degree of uptake particularly for captured probes and the degree of activation for activated probes.
Here in this table are three major categories of probes, the first
being targeted probes. There are a number of examples of these commercially available to today. Then conditionally captured probes, the classic example is there in PET imaging, probes that are uptaken by cells in a dependent fashion and then conditionally activated probes. This is where it gets really interesting and we're getting true molecular imaging through this type of probe. The most common examples are protease‐activated probes through caspases, cathepsins, and MMPs. This is where the FMT comes into play and I'll show you some examples.
Slide 20 Here's the concept of using a probe. So initially, we assume we have
a drug target or a biology of interest. We would then hope to choose an animal model that reciprocates that drug target or biology. So ultimately today and more and more into the future, we want to be able to go to like a kit bag of imaging probes where we can choose a probe that tracks best to that biology or target and also to the model being used. So as a triangle, these three things together become a powerful means for probe‐facilitated imaging of mechanism. That's the concept.
Slide 21 On the FMT, just a little bit on the technology as we get into some
examples here and particularly the CRO perspective. As we choose technologies in any core facility, and I'd argue particularly as a contract research organization, we have to make sure that the technology checks all the major boxes and is going to have great utilization and potential for the industry.
7
So we get quantitative three‐dimensional imaging out of FMT and that's pretty unique for in vivo optical imaging. It's high throughput and efficient. That's needed because we need to be able to run large studies and the probe multiplexing that's made available by FMT provides even more efficient imaging because we're getting multiple readouts in a single experiment. The applications are very broad and they're really only bounded by the probe developers and we're very fortunate that we have developers like the ones at PerkinElmer that are very rapidly providing valuable probes for use with this technology.
It is a true modality for molecular imaging. We're getting down to
disease mechanism and drug response very directly. The applications readily go across disease states and across our therapeutic towers and even into safety so that's important and that's not always a feature of a lot of the more prominent imaging modalities. It has clinical translation potential, which is really the icing on the cake. So it does check all the major boxes.
Slide 22 What we're doing at Molecular Imaging right now is to validate FMT
based biomarkers and our definition of validation is what we call pharma‐ready. What that means is that we're validating it at an extensive level to make sure that the biomarker correlates with disease progression and also response to therapy. That it also correlates with clinical measurements and biomarkers and histopath. We want to make sure that we understand the complexities of study design, how do we get the most out of the technology and ensure that it's quantitative and what are the limitations. Ultimately, we want to make sure we understand the advantage and value of using it versus doing something else.
We have to be able to do this level of validation for every probe or
biomarker we want to use, in every model that we want to use it in, and ideally with each treatment that we want to use it in parallel with. So it becomes a pretty grand undertaking.
Slide 23 Fortunately, some of this is already being done for us. We're not the
first to try to do it. One great example of that is the work that came out of Jeff Peterson's lab at PerkinElmer in collaboration with Pfizer researchers. I encourage everyone to pull this paper. It's a great example of pharma‐ready validation in an RA model; it's a mouse RA model.
8
[0:15:46] Slide 24 Here's the upshot of what was done. On the right, you see the
image‐based biomarkers that are accessed through FMT imaging using ProSense and MMPSense and you see a very tight correlation across a number of treatments in this RA model correlating with the histopath score along the X‐axis. On the left, the traditional forms of measurements through clinical score in paw swelling and you can see a decoupling there. The data points are falling off that straight line. So this is an example where FMT provides a more accurate assessment of disease response in a model like this. So that's demonstrating true value.
Slide 25 How does that look at the ground level? Well, here are some imaging
we did in an RA model of our own and you see we used a ProSense here that is cathepsin activated and so with the inflammation response in the hind paws, you see a 5 to 6 induction of signal increase over time as disease progresses. So that's what we're doing and then we can track a response to a therapy.
Slide 26 It becomes even more interesting when we multiplex and here we're
using a bone‐targeted agent in exactly the same mouse and seeing an increase in the signal level of this bone‐targeted probe that's reading out bone turnover at a later stage in the model as we'd expect the bone degradation in this model. So we're getting two modes of treatment response potentially in a model like this through probe multiplexing.
Slide 27 Another example where we validated the concept is in the mouse
sponge granuloma model. It's a great model for trying to prove concepts.
Slide 28 And just taking a step back, here's an MRI we did in this model and
this is a fluorinated nanoparticle that's taken up selectively by macrophages. You can see the sponges at the top of the mouse there, on the dorsal aspect of the mice here implanted subcutaneously and the one on the right has an inflammatory agent in it so it's promoting an inflammation response. So, it's a really good
9
model for looking at inflammation acutely and you can see a collection of macrophages around the sponge that are activated.
On the left in the PBS control, you don’t see that and there's also
notably some activated macrophages in spleen and liver in the model with the CFA on the right.
Slide 29 So we then took this into use with FMT, with optical probes and we
did some 2D reflectance imaging here with cathepsins and the ProSense agent, and you can see that over time, we've got our PBS sponge on the left and now complete Freund's adjuvant sponge on the right. You do see increasing signal relative to the control over time. So, this is validating the concept of imaging inflammation in a very acute level.
Slide 30 We extended that to MMP through MMPSense and this is the data
summarized on this slide here. But you can see that the diseased animals promote much stronger signals to do with the activation of these probes. So we're getting a very quantitative and acute readout as soon as six hours after introduction of these imaging probes.
Slide 31 The final example I want to show before just ending the talk is to
look a little bit at tumor imaging and this is something that we've been focusing on a lot. What we embarked on initially was a pretty ambitious program where we did perform 104 scans in a day.
Slide 32 to Slide 33 What we wanted to look at was a couple of things. One was a broad
panel of tumors. So we looked at 17 different subcutaneously implanted tumor types and we looked the combination of 5 different optical probes that are used in association with the FMT. Then we secondly wanted to validate the concept of high throughput imaging with FMT. So 104 scans in a day sound ambitious, but it was something we were able to achieve relatively easily.
Slide 34 On this slide, you can see the results and what we've found was in
conclusion FMT technology can indeed be used for tumor tracking and again we're not the first to show this, there's a lot of published evidence of it. But it was nice to be able to do this broadly and quickly. What we found was that the best probe doesn't need to be
10
chosen for each model so you can see a couple of flat lines there where a probe didn’t show progression of the disease with tumor growth. But importantly, the imaging throughput for FMT is not limited so as a CRO or as an industry lab, we can run large, highly powered, industry‐relevant studies, which is an absolute prerequisite for us.
[0:20:25] Big advantage in trying to do this with FMT is that it would not
require luciferase‐transfected cell line because bioluminescence is the more traditional way to do it. I'm not saying it would replace bioluminescence, but there are cases where that would be valuable. It would open up the ability to use primary models for example if we can validate it particularly in deep tissue models where it would be needed.
Slide 35 to Slide 36 Lastly, expanded applications. Here, we're going to expand the
pharma‐ready validation into a number of different disease states. Some of them are shown on this slide in biodistribution. So it's a positive outlook in terms of the breadth of applicability for FMT technology and what we ultimately want to do is provide expanded use and benefit for this technology in pharmaceutical research.
So that ends my talk now. Thanks everyone for listening in and I'll
throw it back to you, Sean. Slide 37 Sean Sanders: Great. Thank you so much Dr. McConville. Very interesting
presentation and great introduction for us. Slide 38 So we're going to move right on to our second speaker for today and
that is Dr. Matthias Nahrendorf. Dr. Nahrendorf is currently an assistant professor at Harvard Medical School and director of the Mouse Imaging Program at the Center for Systems Biology at Massachusetts General Hospital. He completed his joint Ph.D. and M.D. studies at the University of Heidelberg in Germany before moving to the University of Bonn for an internship and then the University of Wurzburg where he did his residency, fellowship, and postdoctoral training. Dr. Nahrendorf joined Harvard Medical School in 2004. His laboratory focuses on the cellular and molecular processes during the healing phase after myocardial infarction using the entire spectrum of imaging modalities, including MRI, nuclear,
11
and optical imaging techniques, with a special interest in multimodal imaging. Dr. Nahrendorf was the recipient of the Helmholtz Prize in 2003 and the Society of Molecular Imaging Young Investigator Award in 2005. Welcome, Dr. Nahrendorf.
Slide 39 Dr. Matthias Nahrendorf: Thank you, Sean. I would like to start off by listing some advantages
that I see in using fluorescence molecular tomography for your research. What we're doing here really is noninvasive sensing of fluorescent molecular agents and also increasingly fluorescent proteins so the mouse doesn’t get hurt. The method is really fully quantitative so we will yield tracer concentrations in 3D volume datasets. We can also start to think about multiplex imaging to assess more than one biomarker. Currently, some of the systems allow up to four channels and we've spent three so far with quite good success.
It's also a modality with high throughput. We can finish up one
channel scanning in five minutes. It's very versatile because we rely on fluorescence so we can combine it with intravital microscopy or fluorescence histology or flow cytometry of liquids or liquefied organs after we did our in vivo imaging.
Slide 40 So there are obviously, as to all imaging modalities, some limitations
and one of the major limitation is that light travels only so far in tissue. That's why we're currently confined to doing this in small animals. The spatial resolution is somewhere between 1 and 2 mm so that's about the range that we get in small animal PET imaging. As in PET imaging, the spatial information is fairly limited because what we're looking at in the FMT dataset is the fluorescence concentration so you don’t get much anatomy with it. We usually try to overcome this by adding an anatomic modality in a hybrid manner.
We have to deal with autofluorescence, as in all of fluorescence
imaging, and there are some tricks to deal with it. For instance, near the mouse we remove the hair or put mice on a non‐fluorescent diet for at least a week prior to your imaging procedure.
Slide 41 Now how does it really work? I'm showing one example here. So
there are some academic centers that even use a rotational gantry where you have comparable to CT imaging, the laser and the CCD
12
camera rotating around the mouse. But this is a setup that we are using right now where the laser moves through 80 positions beneath the mouse, and it's a transillumination setup where the CCD camera then detects the fluorescence and it's positioned above the mouse so you're really in transillumination mode. While this laser is moving through 80 different positions, we're also using transillumination information to take into account that light gets absorbed while travelling through tissue and that there are some scatter.
[0:25:47] These datasets are fed into a reconstruction algorithm that yields a
three‐dimensional map that's show on the right here. You can now put your region or volume of interest into that and it will give you a concentration of your fluorochrome.
Slide 42 If you could see in that map, you have a lot of blobs out there and
you want to find out where is your signal coming from, and we routinely use either CT or MRI to give us this concomitant anatomy. There's quite a neat setup here. You can see that the arrow is pointing to fiducial markers and the top left panels show you the CT image of a mouse that's in an imaging cassette. You little imagine of this cassette on the right upper hand corner. Then the FMT dataset is on the bottom left and on the right, you get an image of a fused dataset where you now can tell where your fluorescence information is coming from.
Slide 43 This has been especially important for imaging small targets. So, we
are interested in imaging atherosclerosis. You're looking at a protease sensor here in apoE knockout mouse and you can see that really the atherosclerotic plaques are very small. So the upper right‐hand corner shows you where you have most of your plaques. That's the aortic root; this is where people usually look by histology. Then right next to that histology, you see an excised aorta, this heat map of protease activity. You see that the root is really hot, but the root is very small. It's just 1 mm in diameter or so. So we use concomitant CT information to really pinpoint our fluorescence information to the root. If you don’t do this, it's harder to tell is your signal coming from the root or is it coming from a lymph node that's somewhere next to it. So this setup is now routinely used. It might not be necessary if you're looking at larger targets such as tumors or implanted tumors where you actually know where your target will be.
Slide 44
13
So, this slide shows you a typical FMT‐CT setup. It actually also includes PET imaging and this is how we do it in our lab. The panel A shows you the nanoparticle platform that we used in this specific experiment. It's a nanoparticle that has a 30‐nanometer diameter and is targeted to inflammatory cells. It's taken up avidly by macrophages, monocytes, and to some extent by neutrophils that has a dextran channel that's derivatized with some handles where we can attach useful ligands such as fluorochromes to detect our nanoparticle by FMT, but also PET isotopes so in this specific example, be clicked on FAT and target ligands, targeting ligands so for instance peptides.
Then panel B shows you the image acquisition and this setup allows
you to basically snap the animal into a holder where you first do your FMT and then you can just go over to the PET‐CT setup, which obviously should be nearby so the mouse doesn’t wake up in the process. But it won't move so you can really then, as shown in C, fuse your channels. The particular images that you see on the right‐hand side show four different channels for fluorescence information so these are the first top four panels and then fused all together with the CT on the lower right panel.
Slide 45 Now how quantitative is FMT imaging? We used our nanoparticle,
which has the advantage that you have a fixed ratio of a PET isotope and a fluorochrome to look into this a bit closer knowing that PET is considered the gold standard of quantitative noninvasive imaging. We used two nanoparticles here, one used 18F, that’s a PET isotope and the other copper 64, which is shown on the bottom. You see an agar phantom that has the dilution curve of our nanoparticles in there and as you would expect PET is more sensitive about an order of magnitude, but FMT is there, it's fairly sensitive. If you look at the correlation curves on the right, it shows you really a high correlation between the fluorochrome concentration and your PET signal.
[0:30:46] Slide 46 So how does that look like in vivo? We implanted tumors here in the
flank left and right side of a mouse and did FMT, CT, and PET imaging in the same animal. So the first two panel shows you the 2D and 3D FMT‐CT information and then the second one PET. So remember, we are looking at the same nanoparticle, but the FMT uses the fluorochrome on the nanoparticle and the PET uses the 18F isotope. The bottom panel shows you fusion of all three modalities and you
14
can see that the PET signal is a little bit larger probably because PET is somewhat more sensitive than FMT imaging.
Then if you just go to the correlation panels in the middle, the lower
one shows you the in vivo correlation of optical tomography with PET imaging and on the right, you see how the co‐registration worked.
Slide 47 So, this panel shows you a more advanced setup where we really
pushed the envelope towards multichannel imaging. So, we did three‐channel FMT‐CT here. We used an integrin‐targeted probe on the left then a cathepsin‐targeted protease sensor, and then you see the macrophage targeted nanoparticle, which is detected by both, by FMT and by PET‐CT. Then on the right, you can see fusion of all of these data. This is interesting because you can now look into a network of biomarkers at the same time, for instance to find out what does your drug do to all of these biomarkers, which are all considered important. Because integrin is for instance involved in angiogenesis, cathepsin protease activity is a prognosis marker, and the macrophages are the source of proteases so you can really start to integrate.
Another aspect is if you're testing a new probe, you might want to
know how it actually works in the environment of all of these biomarkers so you can actually see does it go up if integrin signal goes up and so forth.
Slide 48 So, I have a couple of slides that show you typical applications that
we have recently done in our lab. So, the next one will show you if it comes up. I can't see it on my screen, Sean. This should be the slide showing endocarditis. I hope you can advance to that slide. Yup, now I see it.
Sean Sanders: Yup, I see it. Dr. Matthias Nahrendorf: Yeah. So what we did here is we used FMT as a modality to really
pursue probe development before we got into the more expensive and more time consuming PET imaging. So, we developed a probe here to detect infective endocarditis and the end goal is to really use this in patients because we are lacking good tools to detect infection of the heart valves. Right now, we're relying on fairly unspecific criteria such as a new heart murmur or blood cultures, which reflects
15
on circulating bacteria, but it doesn't tell us about the site of infection.
So, the left upper‐hand side shows you the model that we used here.
So, we had inserted a catheter into a mouse heart and then injected Staph aureus and it formed the typical vegetation shown on the histology in the middle top panel. The probe that we designed is shown on the right. It uses and engineered prothrombin analogue that we first inactivated to avoid clotting in the blood stream and then derivatized this fluorochrome and later with a PET isotope, and it binds really tightly to staphylocoagulase. Staphylocoagulase is secreted by Staph aureus and it's a virulent factor so what we can do now here is in vivo black culture so to speak and detect the presence of bacteria in vegetations.
[0:35:38] The FMT data is shown on the lower left. You see Staph aureus, a
nice signal there in the left ventricular outflow tract that if you infect the mouse with a different bug, you don’t get any signal. This encouraged us to then go on and develop the PET tracer.
Slide 49 So, you can really use a modality that's very efficient and high
throughput to do proof of principle and then move on to develop a tracer that can go into clinical translation. I also see a role for FMT in drug development. So Pat already talked about efficacy studies and I'm showing here one example also based on our endocarditis work. So what we are doing here is we are using a drug that is known to be effective to kill S. aureus vegetations and you can see that we can actually follow this by FMT imaging.
So on the left‐hand side, again a mouse infected with S. aureus and
on the right‐hand side a mouse that was treated with antibiotics. On the lower left panel, it shows you that you now get a fluorescence readout in picomole that tells you vancomycin is killing off the bacteria and my imaging target is diminished. On the right‐hand side, I thought it's interesting to show that these mice also survived better when they're treated, but if you stop the treatment, that's indicated by therapy end, there's actually a recurrence of disease and that's very close to what we see in the clinic.
Slide 50 I have another example here up next that shows you how FMT
imaging could be used in drug development. It features our work in
16
developing nanoparticles for delivery of siRNA. So what we did here is we thought of a concept to deliver siRNA into inflammatory monocytes to silence the CCR2 receptor and that receptor is responsible for recruitment of monocytes into sites of inflammation. So we see this as a new concept to curb inflammation and to do so by just specifically targeting pro‐inflammatory leukocytes.
Slide 51 We used FMT imaging to follow the biodistribution of our delivery
vehicle. So what we did here is we labeled our siRNA with near‐infrared fluorochrome and dynamically imaged the mouse over time. The upper panel shows you the mouse five minutes up to two hours after injection of these nanoparticles and you see on the left‐hand side that there's nice blood pool signal, and on the right‐hand side you see that there's quite a bit of signal in the spleen. Again, you can really quantify this over time. It gives you a nice time curve. We even fitted the blood pool signal and got a blood half‐life of eight minutes for our nanoparticles.
Of course, then you can follow up with fluorescence imaging shown
in histology of the spleen on the right and that shows you that the siRNA is indeed CD11b positive monocytes in the spleen.
Slide 52 So, my last slide shows you a basic science example of how we use
FMT imaging. One big advantage is really that you don’t have to kill the mouse to get a molecular readout. So we designed a longitudinal study here where we induced myocardial infarction and then did two MR scans, one very early on to get the infarct size and the size of the heart and the function of the heart very early on. Then we did FMT‐MRI for our molecular markers. In these particular experiments, we used nanoparticles targeted to macrophages and also a protease marker. Then since we didn’t have to kill the mouse, we could relate our molecular findings to the follow‐up MRI three weeks after myocardial infarction.
We looked at two specific cohorts here to figure out what impact the
splenic monocyte reservoir has on infarct healing. So if you look at the FMT‐MR images that are on the left‐hand side in the middle, you see that if you take away the spleen from these animals, the nanoparticle signal and the ProSense signal is dropping. Then because we don’t have to kill the mouse, we get this noninvasively, this signal, we can follow up three weeks later and find that this drop in molecular biomarker leads to the drop in ejection fraction so we
17
see this here because you take away the necessary splenic monocyte reservoir.
[0:40:43] Slide 53 With this, I would like to conclude and also acknowledge my
collaborators who all use FMT and have fun doing so in their studies. Sean Sanders: Great. Thank you so much, Dr. Nahrendorf, for a very interesting
presentation and some great work there. Slide 54 So, we're going to move right on to our Q&A. Just a quick reminder
to those watching us live, you can still submit questions, just type them into the box on the bottom left of your viewing console and click submit. If you don’t see this box, just click the red Q&A icon and it should appear.
So, Dr. Nahrendorf, I know you talked a little bit at the start of your
talk about background fluorescence, but we have a question here that asks if you can speak a little bit more about that. What are some of the fluorescence problems, the background fluorescence problems that you encounter and how have they been solved? And since you'd spoken about it already maybe we'll go to Dr. McConville first and see what he has to say.
Dr. Patrick McConville: Yeah, okay. Thanks, Sean. That's a great question. Fortunately, it's a
big challenge when we're looking at particularly systemic tumor models and deep tissue tumor models because autofluorescence is a particular problem in the gut and abdomen. Fortunately, there are a few simple things you can do to minimize it through the use of, as Matthias mentioned, chlorophyll‐free chow a week in advance at least of imaging. We can do things like fast the animals overnight similar to what we do with PET imaging and we can use things, like there's a product called GoLYTELY that will clear the GI region. So, the folks at PerkinElmer have good experience with this and they're a great resource for letting you know about products and procedures for doing that. But those will be my comments. What about you, Matthias?
Dr. Matthias Nahrendorf: Yes. So what we do routinely is to remove the hair from the mice just
before imaging so that's an important step, and I think going to the near‐infrared for imaging already diminishes autofluorescence. So you won't really get rid of it altogether, but I think, you know,
18
imaging in the near‐infrared where FMT is happening deals with this problem quite nicely. We won't be able to completely get rid of it, but if you're aware of it especially in applications where your imaging signal is fairly low, you can take this into account during analysis.
Sean Sanders: Excellent. I'm going to stay with you, Dr. Nahrendorf, with a question
on resolution and this viewer asks whether individual cells can be seen in live animals using this technology.
Dr. Matthias Nahrendorf: So I think that won't be possible. What we can do is we could look at
population dynamics. So if you want to look at single cell resolution, what you have to do is do intravital microscope. It's obviously not as noninvasive, it will have to get very close to the tissue and your penetration depth is limited, but there you can really see individual cells. But I think FMT has a pretty good capability of looking into populations and even doing this in a quantitative manner. I think that this will even improve with increased availability of reporter proteins in the near‐infrared such as mCherry. So for instance if your cell population of interest expresses a reporter that you can pick up as FMT, you can follow the population dynamics over time or in response to a therapy.
Sean Sanders: Excellent. I have one for you, Dr. McConville. What is your opinion
on bioluminescence imaging of luciferase‐expressing cells to quantify cell retention after transplantation in infarcted myocardia? Actually, maybe this is for you, Dr. Nahrendorf.
Dr. Matthias Nahrendorf: So, I think bioluminescence is good. It has really a lot of advantages.
It's very sensitive and it has very low background. Where it really differs is that it's not fully quantitative. So you're looking at photon counts and depending on where your imaging signal is located, you know, something that's close to the surface will appear brighter, something that's deep inside the mouse is not as bright just because of effluxion. FMT controls for this and I know that there are groups out there that are trying to build tomographs for bioluminescence imaging but at this point, if I wanted to do this kind of thing, I think there are some limits with respect to quantitation.
[0:45:45] Having said that, I think if you're comparing cohorts where you know
that the depth of your imaging signal will be about the same or if you want to follow mice over time, bioluminescence is really a good modality.
19
Sean Sanders: And now one for you, Dr. McConville. In terms of using FMT for PK profiling so looking at different antibody PK comparisons, how do you overcome the issue of labeling heterogeneity?
Dr. Patrick McConville: As an imaging provider, that's something that we really rely on
external providers to overcome. There are a lot of kits out there now from the commercial entities to provide pretty routine and standardized labeling, but it's definitely a challenge. And these fluorophores are quite large generally so anytime we're bringing that on board even with a biologic, we've got to be prepared to assume that there could be some change to the kinetics of the antibody or the protein that's being labeled. So it's not a challenge that we take on here, but it's definitely a good question and in doing anything, any type of labeling of the molecule when we're looking at biodistribution we have to consider the effect of what the labeling is doing.
Sean Sanders: Dr. Nahrendorf, do you have any comments on heterogeneity of
labeling efficiencies? Dr. Matthias Nahrendorf: It's a very important question because what you're measuring is the
concentration of a fluorochrome. So if, you know, one batch of antibody has twice as much fluorochrome than the other batch that's obviously a problem and you can't compare it. So good QC of your probe is important and that you really need to do if you want to image cohorts and be comparative.
Sean Sanders: Excellent. So this question might be able to be answered by both of
you. I'm sure you both have some inputs. The viewer asks whether tumor progression can be monitored live over a period of time in one particular animal that you keep alive without the need to sacrifice animals at every study time point. So, Dr. Nahrendorf, you'd like to start us off?
Dr. Matthias Nahrendorf: So it really depends. I think you can do this with FMT. The best
approach that right now is probably cutting edge and not really that far, not that easily available would be if your tumor cells express protein in the near‐infrared. What you can do more easily is use a probe that targets something in the tumor microenvironment or the tumor cells or maybe the tumor vasculature that will give you some sort of signal, but then you're really also subject to changes in the tumor microenvironment. If the tumor for instance gets less vascularized over time that may change your signal.
20
Sean Sanders: Dr. McConville? Dr. Patrick McConville: Yeah. I would agree with everything Matthias said and I'd just add
that FMT it can be done even now, as he said, through activated probe use and there are a number of deactivated probes that can do this potentially because tumors fortunately do express things like MMPs and cathepsins and those types of molecules. But the advantage of doing it with FMT versus an engineered cell line is a huge saving potentially in the time required to develop an engineered cell line that expresses a near‐infrared moiety or luciferase and then not having to worry about the changes that occur in transfecting the cell with that reporter.
Sean Sanders: So I'm guessing that you could use this procedure for longitudinal
studies for tumor development and response to antitumor treatment?
Dr. Patrick McConville: Yes, yes. Sean Sanders: Excellent. So for you, Dr. Nahrendorf, for FMT‐CT, how do you
exploit anatomical information to improve your 3D modeling of the FMT?
[0:50:02] Dr. Matthias Nahrendorf: Well, I think so groups are working on this and building integrated
scanners to really use prior anatomy information to improve their FMT reconstruction. I guess this is what the question is aiming at. Similar to what you do in PET imaging where you do an attenuation correction scan by CT, this will most likely improve the spatial resolution of FMT imaging so that is something that in the future will improve spatial resolution and maybe also sensitivity.
Sean Sanders: Could you talk a little bit more, this viewer is asking about the major
differences between bioluminescent imaging and FMT and IR imaging and which might be better suited for deep tissues like pancreas and the liver, etc? So, Dr. Nahrendorf, you'll start us on that one.
Dr. Matthias Nahrendorf: Yeah. So I mean both technologies or both modalities are somewhat
related, you're looking at photons traveling to tissue, but there are also some important differences. So in bioluminescence, you don’t excite the tissue with a laser. So if you think about the firefly, it just glows, you don’t need a laser to make it glow and something similar happens in the mouse whereas in FMT, you need a laser to excite
21
your fluorochrome. Another difference is that in FMT imaging, you're further to the near‐infrared so the wavelength is further out. Bioluminescence is something your emission wavelength is maybe in the 560 and FMT is using something like 680 or so forth so you should actually get a little less absorption.
On the other hand, bioluminescence imaging is pretty sensitive and
has a very low background because you don’t use a laser to ‐‐ and using a laser creates more autofluorescence. The biggest difference really from my point of view is that FMT is quantitative. It really quantitates a concentration of fluorochromes and that's really helpful if you're looking at comparing cohorts. Whereas bioluminescence imaging is something that is really nice to do to follow reporter genes over time and this is usually expressed by either cancer cells or the stem cells are also a frequent application. So, I would say that while there are some similarities, there's really a big difference between both modalities, the major one also being the ability to do total quantification of the signal.
Sean Sanders: Dr. McConville, anything to add? Dr. Patrick McConville: Yeah. The background issue is particularly important when we're
looking at systemic models like intravenously injected leukemia cells that are populating the body or multiple myeloma models that are similar then bioluminescence can be very valuable for being able to quantify that without having potential background that is improved in the near‐infrared. In the end, both can be shown to correlate with tumor growth so if we can correlate either one with tumor growth, we can at least argue that the biomarker is quantitative. It is enabling us to make a decision or to determine the treatment effect. So, I don’t think bioluminescence is going away any time soon or probably never in the oncology realm, but the two definitely have their relative advantages and disadvantages.
Sean Sanders: Great. Now is intracranial FMT imaging possible? This viewer asks if
the striatum or substantia nigra is possible to resolve. Dr. Nahrendorf, you might be able to best answer that.
Dr. Matthias Nahrendorf: It's possible. So you can image in the brain, this has been done. It
depends obviously on how strong your signal is, but if you have a decent enough fluorochrome concentration, it should work.
Sean Sanders: Great. I mentioned at the start of the talk that this technology or at
least some parts of it have been translated into the clinic or they're
22
looking at that. How would this be able to be done considering the penetration issue with humans, Dr. Nahrendorf?
Dr. Matthias Nahrendorf: So people are looking at tissues that you can squeeze enough so you
can shine the light through and where penetration depth is not as big of an issue. So, one idea is to look at the hand in rheumatoid arthritis. Other people are working on detecting breast cancer by tomography or you could think about structures in tissues imaging targets that are superficial so there are thoughts about the carotid artery for instance. But for deeper tissues let's say coronary arteries, it will be very difficult.
[0:55:31] Sean Sanders: Dr. McConville, and what do you see on your side? Dr. Patrick McConville: Very similar I would say to what Matthias had elucidated there. Sean Sanders: Okay. Excellent. So this one's for you, Dr. McConville. The viewer
says that FMT as it appeared in the second talk or appeared from the second talk that it has limited angle tomography. Does this have an effect on quantitative accuracy and also are measurements taken from the images expressed in absolute or relative quantitative sense?
Dr. Patrick McConville: Okay. Yeah, I think in the end there are limitations of course, but the
proof is in the pudding, if you will. So I mean Matthias showed the correlation between PET based concentration of probes and FMT so the evidence is that it is quantitative despite some of those limitations. That would be my comment.
Sean Sanders: Anything to add, Dr. Nahrendorf? Dr. Matthias Nahrendorf: Yeah, I agree with this. I mean obviously if you have rotational gantry
that might improve in the spatial resolution in the Z direction, but in our hands, it works pretty nice even though you don’t really do a rotation of the mouse.
Sean Sanders: Okay. Great. Probably a quick question, the viewer asks to what
extent the animal model itself impacts the outcome of the FMT work and the resolution and images that you obtain, Dr. McConville?
Dr. Patrick McConville: Very much so. So the model itself will really drive the success and we
have to take into account a number of different features of the model. So, you know, that's where the choice of probe and coupling
23
that to the biology of interest that's inherent to that model is important. So, you know, we would start with a concept and a hypothesis that a particular probe or a set of probes should work for tracking some aspect of disease in some models. But then it's up to us to correlate the readout that we're getting through the FMT based biomarker with what we're intending it to read out, which is the disease progression and then secondarily to that, the response to a therapy which is another variable itself. So, yeah, it's absolutely critical.
Sean Sanders: Dr. Nahrendorf? Dr. Matthias Nahrendorf: I agree. I mean we have to tweak our experiment in a way to really
make sure that imaging works well. So for instance, when we started our work on atherosclerosis, one trick that we used is make sure that we image our mice at a time point that we know that inflammation peaks. So I think these kinds of considerations are really decisive for the success of the imaging experiment.
Sean Sanders: Great. Well we are at the end of our hour so I'm going to put out a
final question to you both. One that I like to give all my speakers in these webinars is where do you see the technology moving in the next 5 to 10 years or where would you like it to be so that it can really drive your research? Let's start with Dr. Nahrendorf.
Dr. Matthias Nahrendorf: So I think we are really seeing a lot of improvement already in terms
of throughput so it's very quick. What I like is that we have a lot of channels. We have done three‐channel, four‐channel imaging so this is something I'm excited about and integration with other modalities. I think where we can expect a lot of advance is on the probe side and the reporter protein side, I think that we will see a lot of progress there. With respect to technology, I think where we can always push the envelope and I'm sure that that will happen through integration of for instance anatomic progress in reconstruction that you get higher resolution and even higher sensitivity. So, I'm never satisfied with any resolution or any sensitivity I get so I'm hoping that we will see progress there.
Sean Sanders: And Dr. McConville? Dr. Patrick McConville: Yeah. From my perspective, it's all about the applications and
validating those applications so I'm greatly looking forward to huge inroads with expanding probe portfolios. That’s happening at PerkinElmer and some other providers as well. And just expanding
24
the use into other disease areas and other types of models I think is what the future is going to show with more and more validation that goes on. Then on the technology side, I would say there's a lot of potential for improvement, particularly on the software side and maybe on the hardware so that we could do things like remove background more easily or compensate for that. Maybe even get into things like simultaneous imaging of multiplex probes versus sequential imaging. So those sort of things will drive I think throughput improvements as Matthias said and sensitivity improvements and ultimately better applications.
[1:00:44] Sean Sanders: Fantastic. Well, many, many thanks to both of you for providing us
with such fascinating talks and for answering all the questions from our online viewers. Dr. Patrick McConville from Molecular Imaging and Dr. Matthias Nahrendorf from the Harvard Medical School.
Many thanks to our online audience for all of those questions you
submitted. I'm sorry we didn't have a chance to get to all of them. Slide 55 Please go to the URL that I'll be putting up now in your slide viewer
to learn more about resources related to today’s discussion, and look out for more webinars from Science available at www.sciencemag.org/webinar. This particular webinar will be made available to view again as an on‐demand presentation within approximately 48 hours from now.
We'd love to hear what you thought of the webinar, just send us an
email at the address now up in your slide viewer; [email protected]. Again, thank you to our panel and to PerkinElmer for their kind
sponsorship of today's educational seminar. Goodbye. [1:01:49] End of Audio