In my last post I discussed some of the more standard ways in which ophthalmologists visualize the retina. Today we are going to discuss how Magnetic Resonance Imaging (MRI) is advancing the field. Recently we invited Bruce Berkowitz to give a seminar to our department and I got to talk with him over lunch. Dr. Berkowitz is leading the way in the application of MRI to study and ultimately diagnose retinopathies before it is too late. You may remember, at the end of my last post, I mentioned that we need a biomarker which has the power to predict whether or not retinopathy is progressing in a diabetic patient. Dr. Berkowitz is making great strides towards this goal and he is using MRI to get there.
Like OCT (see my last post), MRI is one of these computationally intense sensing modalities. The science behind it is complicated. Basically, it relies on the ability of water to orient in a magnetic field. Water, H2O, has a concentration of negative charge near the oxygen because the massive nucleus of the oxygen atom pulls the electrons away from the attached hydrogen atoms. In turn, the hydrogen atoms have some positive charge. This is what allows water to spin and orient in a magnetic field. Now, consider yourself laying there in an MRI machine. The magnetic field has just forced all of your water molecules to face in one direction. This takes some energy and that energy is now stored in the water. Suddenly, the field is shut down and the water molecules can relax. Now for the critical part: as they relax they release a photon. It is this light that the MRI detector detects and uses to create the image. How to deconvolute the information these photons provide is far beyond me. Each photon has a wavelength, a frequency, and a direction of travel. Somehow this is enough to create the 3 dimensional images that have proven so important to modern medicine.
Let’s take this one step further. If we introduce some sort of molecule, say – a dye that also orients in a magnetic field and releases photons and this molecule does not distribute randomly in the body, but rather, distributes to specific places – we can increase the contrast of our image and see special features of the body such as blood vessels or tumors. If this molecule does something special like react with a particular component of the body and change its state based on that reaction, why then we can use this as a molecular measurement tool.
Applying MRI to diabetic retinopathy, we know that the blood vessels in the retina are leaky when retinopathy progresses to the clinically debilitating stage. How is this visualized using MRI? One can use a dye that is easily visualized because of its sensitivity to orient in a magnetic field. The Berkowitz group took MRI pictures before and after the dye was injected into the blood stream. Leaky blood vessels were expected to produce a different kind of signal as compared to healthy blood vessels and indeed they did. This is exciting for a couple of reasons. First of all, preliminary work suggests that this may be more sensitive than some of the other techniques. Indeed the success rate of identifying patients with retinopathy using MRI (78%) was superior to that using more standard techniques (65%). At this very moment studies are underway to see just how sensitive this new technique may be.
In addition to detecting leaky blood vessels, the Berkowitz group is investigating retinal oxygenation. The retina is the most metabolically active tissue in the body. It is constantly detecting photons, performing a bit of preprocessing to enhance boundaries and distinguish other features of the visual world, and then sending it out to the brain for further processing. Returning to the factory floor analogy of metabolism that I employed in a previous blog, oxygen is the factory’s power source. This power source needs to be carefully controlled. Too little oxygen and the tissue starves. Too much oxygen and the tissue begins to produce reactive byproducts that cause all sorts of structural damage. All of this suggests that there is a delicate balance that must be maintained between the needs of the retina and the blood flow to the retina. This leads to an important question: can we use retinal oxygenation as a way to predict retinopathy. Keep in mind, we do not know who will develop this complication and by the time we see leaky blood vessels it is too late to provide an effective treatment to block disease progression. The damage has been done. Dr. Berkowitz is proposing that if we can test how the retina responds to a change in oxygen we might be able to detect that subtle molecular signature that indicates a problem before it manifests as retinopathy.
The tricky part is actually measuring oxygen. One cannot simply poke some sort of sensor into the eye and get a reading. As it turns out, however, molecular oxygen has some ability to orient in a magnetic field. This means that if you tune an MRI machine in the appropriate way you can actually see molecular oxygen. Apparently, because of all sorts of confounding signals from other molecular entities in the tissues lying just behind the retina, the best results were obtained by focusing the detectors on a plane just in front of the retina within the vitreous humor. So, what they are actually measuring is the oxygen that is diffusing out of the retina into the adjacent tissue. At any rate, patients were challenged by breathing 100% oxygen. In a healthy patient there should be no change to the amount of oxygen in the retina. The hypothesis was that a difference would be seen in diabetic patients who were having trouble regulating retinal blood flow. The experiment was successful. Berkowitz’s group found that they could, on average, distinguish a difference in response in healthy volunteers as opposed to diabetic volunteers. Importantly, there was no change in this response due to age. As you might imagine this would completely confound things. Another interesting and possibly important observation was that in diabetic patients the difference in oxygen response was variable across the retina. We know that the microvascular abnormalities seen in the retinas of diabetic animals and people are not uniform. The irregularities are often found in just one or a few regions. This suggests that perhaps the differences in the oxygen response seen in different parts of the retina reflect the underlying pathology as it develops.
As it stands, we cannot yet make the statement that this technique will predict retinopathy in diabetic patients. This will require a full clinical trial in which numerous patients are screened using this technique and then followed for years to see who gets diabetic retinopathy and who does not. Just as with cancer, if we can detect the problem early there are a number of possible ways to slow down or even stop the pathology. I’ll describe some of these new therapeutics another time.