Positron Emission Tomography (PET Scan) in the Diagnosis and Assessment of Connective Tissue Disorders

Grand Rounds

Stanley J. Goldsmith, MD
Director of Nuclear Medicine
NewYork-Presbyterian Hospital
  1. Defining PET, its Importance and Value
  2. Positron-Emitting Radionuclides
  3. Imaging Patients with Fluorine-18 labeled Deoxyglucose (FDG)
  4. New Equipment and Buildings
  5. Applications in Common Tumors
  6. FDG in Detecting and Following Inflammatory Disease
  7. FDG in Atherosclerosis Imaging
  8. FDG to Evaluate Loosening Prostheses vs. Infection
  9. Question and Answer Session

Steven A. Paget, MD: It is a pleasure today to introduce a good friend and a respected colleague, Stanley Goldsmith, who is a Professor of both Medicine and Radiology and Director of the Division of Nuclear Medicine at the NewYork-Presbyterian Medical Center and at the Weill Cornell College of Medicine. Stanley, in his many years of interest in endocrinology and working with Nobel Prize winners, has also done extraordinary work in many areas of radiology and nuclear medicine, including atherosclerosis and cancer. He will talk to us today about PET scan imaging with FDG - radiopharmaceutical fluorodeoxyglucose. This type of imaging, to me, is absolutely vital to defining both involvement in certain organs or certain components of the body and also the severity of the disease. Many of our inflammatory processes do not give us excellent ideas as to what the activity of the disease is or whether we've won the whole war. It is a pleasure to have you here today, Stanley.

Stanley Goldsmith, MD: Thank you Steve. Good morning, and thank you for the nice introduction. Some of you may remember that I was here about 18 months ago. We had been referred a case by one of the Hematologists -- a patient without anatomic evidence of lymphoma who had presented with persistent low-grade fevers. Along the way, because we had been imaging a number of patients with lymphoma, we performed an image, and I was struck by the intense uptake in the aorta and suggested that there might be a very active process going on. I suggested atherosclerosis. That led to a work-up and the patient was diagnosed with temporal arteritis in this institution. I think Dr. Laskin probably saw the patient. I should have brought that along, but we had more exciting things to talk about. At that time, I reviewed PET overall. For the younger people in the audience who weren't here, I thought I would review some of the basic material. And for the older people who were here, they have short-term memory gaps, so I thought it was worth a repeat anyway. I'd like also to tell you what is going to be happening at this medical campus within the next few months, which will make this exciting technique more available to you for both clinical applications and investigation.

Defining PET, its Importance and Value

So we will begin with what is PET and why is it important and what is so special about it. There are really four points which I will review in the next slides, but briefly: What is a positron emission annihilation? What is so special about coincident imaging? What is so special about the radioactive isotopes, which is what we call the radionuclides that emit positrons? And why do you need a cyclotron? So with that we will begin on the left.

I have illustrated probably the most commonly used positron emitting radionuclide, chlorine-18, over here, which decays by the emission of a positron. We won't go through all the things that influence that. It is one of the ways that a radioactive atom becomes stable. It gives up a particle. In this case, it gives up a positron - that science fiction particle - which has the mass of an electron but an equal and opposite charge and, hence, it's called a positron.

Our universe is full of electrons in all of the atoms in our body and the universe. So, shortly after emission, even if it is just fluorine and water, it will interact with one of the abundant electrons. Electrons, if they come from the nucleus, are called beta particles and designated beta minus instead of beta plus. These two, like the black and white Scotty dog magnets, attract each other and annihilate, meaning that their mass is converted into energy. The energy is released in the form of two photons - just like light photons, X-ray photons, gamma photons - with a characteristic energy. The energy is 5011 KEV. That is more energetic than the typical diagnostic nuclear, and it's more energetic than the typical X-ray. It doesn't mean much. It just means that it is more difficult for us to capture it with instruments. We need a bit more shielding, thicker crystals, and sometimes other material to detect it. But a nice feature is that the energy equivalent of this mass is described by the famous Einstein formula E=MC2. These photons come off in 180 degree coincidence from each other. This is the only instance where that happens. In all the other forms of radioactive decay, a photon comes off, another photon comes off - it's random in a 360 degree solid angle. Now these come off together in a random fashion 360 degree solid angle, but always in a pair and also coincident.

As a result, the many things that we have to do with nuclear medicine instruments to figure out where they came from, we don't have to do. This is a typical PET scan. This is the type of scan which they have at Memorial, which they have at East River, and which we've ordered, but we since have upgraded it. This ring detector looks very much like a CT, which has all detectors around the CT. An X-ray tube passes through the patient, and the information is recorded around, and it's a back projection reconstruction performed. In nuclear medicine, we don't use an X-ray tube. The signal - the photon - is emitted and detected. We can have a ring detector with non-positron emitting radionuclides. But when we have a positron emitting radionuclide with the coincidence photons, we have a system which is inherently tomographic, because when they come off there is always a ray. We don't have to go through a complicated positioning algorithm and say, "Where did it come from?" We know it came from this path, and since we have detectors all around, it's getting the one that comes this way and this way. It is quite an electronic feat to determine that this and this occurred at the same time, but these electronics can do that within fractions of seconds. Very small fractions of a second. It will define the ray, and by putting them all together with the computers that are available now, we can reconstruct where the information came from. So it is very similar to CT, MRI, back projection reconstruction, based on the fact that the positron emits a coincident signal. I hope that is clear.

Positron-Emitting Radionuclides

Now, moving on to the positron-emitting radionuclides, this is particularly, doubly, fortunate because the atoms of life oxygen and nitrogen (are positron emitting), but, unfortunately, hydrogen is not. Let me pause for a minute and discuss that. The hydrocarbon chemistry is mimicked reasonably well by fluorocarbon chemistry. In the early days of nuclear medicine, we began working with radioactive iodine. That wasn't very good for the thyroid. But then we had to learn to label other things with the iodine. More recently, we've replaced the iodine with other radioactive isotopes that have more favorable signals. But it is always a pretty unnatural compound.

When you can label one of the carbons in a glucose molecule with a radioactive atom, you have a very negative compound. When you can label oxygen or nitrogen with a radioactive oxygen and nitrogen, you do very well. There is no gamma-emitting or positron-emitting form of hydrogen, but fluorine behaves more like hydrogen in organic chemistry than it does like a halogen. It is a very tight bond, and in biologic systems, fluorinated compounds retained their activity. Indeed, as you think for a minute from your own experience, many drugs which are of the fluorinated form have increased activity because they are recognized by the receptors. And, if anything, they don't dissociate, so fluorinated steroids will have increased potency over the negative steroid molecule.

So substituting the fluorine for the hydrogen is not as awkward a maneuver to do as iodinating a compound or trying to get a technetium label on or some other label. So from that, although we and others go through proofs that the fluorinated form is behaving like the native molecule, for the most part that is true. So you can make fluorodeoxyglucose that behaves like the oxyglucose would, which behaves initially very much like glucose. You can't make iodinated glucose and get it to behave like glucose.

You can label many neuro-ligands and other compounds with fluorine-18, and they mimic to a great degree the nature ligand, dopamine and so on. The form of carbon that is available as the positron emitter is carbon-11, but one of the problems is that it has a 20-minute half-life, which means that half of it disappears every 20 minutes. Nitrogen-13, 10-minute half- life. Oxygen-15, two-minute half-life. The half-life of fluorine-18 is 110 minutes -- almost two hours. We used to think of that as very short. It's now considered almost a leisure-time activity. You don't need a cyclotron to make this. You have a cyclotron, in two hours they can make enough of it and ship it to you on demand. It's like buying ice: you bring it home and you lose some, but it's available.

Imaging Patients with Fluorine-18 labeled Deoxyglucose (FDG)

We are daily imaging patients with fluorine-18 labeled deoxyglucose, with some of it made in an industrial cyclotron house on the Columbia campus. Other material has been shipped to us from outside of Baltimore, and commercial cyclotrons have been built to do this in New Jersey. And, in addition, Cornell has invested in a joint project with Memorial in a cyclotron which is in place on 72nd Street. The building is being completed around it, and it will be installed in this quarter. We hope to have it operating by the end of the year. Now, whether we can convince the chemists to get up at 4 a.m. to make the materials so we can injected it at 7 or 8 a.m. is another story, so we may have a cyclotron and still be buying glucose at least until noon. They are very rapid syntheses.

So this is available commercially, and everybody is doing it. And the FDG (radiopharmaceutical fluorodeoxyglucose) is approved for some clinical applications, which I will review in a moment. These can only be made by a resource that has a cyclotron, and that is the basis for the Dean's Office at Cornell supporting the purchase and siting of a cyclotron, particularly the oxygen-15 water. Oxygen-15, interestingly enough, is incorporated in water, and that has become the method for radionuclide flow studies. And you can make enough of it, inject it, and get good signals with very favorable radiation dosing to the patient. So much so that IRB committees will allow you to do a dozen or two dozen studies on volunteers with oxygen-15 labeled water a year, whereas they will allow only two studies for research purposes with fluorine-18 FDG. There are different criteria if you are doing a study in a volunteer as opposed to a clinical procedure in a patient. You can do as many of these as you need to do for clinical approved indications. But to study volunteers, there are very severe limitations, and despite that very dramatic term "annihilation" and E=MC2, the radiation dose from all this to the patient is very favorable and comparable to or less than other diagnostic procedures.

The nitrogen-13 has been used a little bit to make ammonium ion, which is used as a perfusion agent. It rapidly crosses membranes. Carbon-11 is the most interesting compound and the one looked to for the future for a lot of ligand work to study neurochemistry, the chemistry of inflammation, and other processes. But whenever someone finds a carbon-11 compound of interest, other people will pursue whether the fluorine-18 compound will give similar results, because it is much easier to work with a 110-minute compound than a two-minute compound.

On your right, I have illustrated what goes on with fluorine with FDG. Some of you may or may not know, now that the f-18 I have already indicated serves as the marker. It binds very tightly to the carbon and will stay on very well. We need the deoxyglucose, as opposed to the glucose, to keep the whole molecule from breaking down after it enters the cell. So it enters the cell by the usual glucose transporter mechanisms, which express themselves in every cell. And after that though, after it undergoes phosphorylation in the glucose-6 phosphate - in that case deoxyglucose-6 phosphate - it doesn't go on to the other steps of glucose metabolism. It stays. If it did, it would move all over. Likewise, even when carbon-11 is used, they use carbon-11 deoxyglucose. Using glucose would give you a cloud of radioactive carbon dioxide.

Now believe it or not, it works. Very sophisticated work. I excluded from the review images of the myocardium, where the primary substrate of the myocardium is free fatty acids, where under conditions of low oxygen, ischemic myocardium, it shifts to glucose utilization. You can see that in a patient at rest versus during stress. That is used to identify what is called viable myocardium, chronically ischemic viable myocardium, which cannot be differentiated from scar by any other technique except demonstrating that there is glucose metabolism. In this noncontracting, nonresponsive, and apparently nonperfused myocardium, there is a degree of glucose metabolism when it's viable as opposed to when it's scar.

New Equipment and Buildings

Now, what is going to be going on in our neighborhood? I've already referred to the 72nd Street building, which is a one-story building going up behind the S-Building, right next to Sotheby's. It's one story above ground and two stories below ground. At the lowest level, chipped out of bedrock, is a vault containing cyclotron and radiochemistry labs. The cyclotron is jointly operated for us and for Memorial. There are separate Cornell labs and Memorial labs. The primary commitment of this facility is research; so we have been talking with Dr. Lockshin, particularly, about potential research projects. But you are all invited to discuss or think about other applications, and this would be available to you as members of the Cornell faculty.

We've already ordered a high-end PET scanner, like the one that has been in use at Memorial for a few years. As I showed you, we have been trying to convince the Dean's Office to upgrade it to this instrument, which will include a helical CT, and permit us to take this sophisticated CT, which has this beautiful anatomic detail image, and marry it to the PET. This is an FDG image, a cross-section (the chest of a patient with a tumor in the chest wall) and to fuse the images - because they are done on a single instrument, on a patient on the same couch without getting up, one after the other in sequence. And I believe that this instrument is going to revolutionize both nuclear medicine and radiology, and certainly the practice of oncology. It is going to get radiologists interested in positron emission if they don't have to figure out how to read this. It's going to take the CT that they know how to do and color the tumors in red. So anyone can do it.

In addition, we have ordered one of these for the Clinical Nuclear Medicine Department at the hospital, but siting it in a house in New York State is much more difficult than buying an empty supermarket and putting it in. So later this morning, we are going to the final planning session with the architects about what has to be done. We expect that it will be up in early 2002. This should be available in January.

Now, in the meantime, the images that we did for you last year, we did on a modified gamma camera, a dual detector system. That system we had removed about two weeks ago. We are currently having installed an improved version, which will have a weak CT in it; so we will be able to do fusion imaging. It won't have the anatomic resolution of this instrument. So we will be back on the air for clinical studies with that instrument, I hope within two weeks and, as I said earlier, this high-end instrument plus the research instrument.

Now, what is this good for, and what has it been used for? The approvals by HCFA for reimbursement have been limited primarily to oncology. They have under review applications for diagnosis of Alzheimer's, seizure disorders, and cardiology, but for the most part, the neuropsychic imaging, which is really where this was first developed extensively, is still limited to research applications. Also, cardiology is primarily research applications, and there is growing interest in other applications - inflammation, metabolism, and so on. There is really no limit to what can be done with it. I guess the challenge, in addition to figuring out what to do, is how to fund it when it is not reimbursable.

Applications in Common Tumors

We can spend many hours just on the use of FDG imaging in oncology. These are four of the tumors which have been approved for some time - application in carcinoma 1 for diagnosis, now staging recurrence. Sensitivity for detection of disease compared to surgical follow-up CT, which is viewed as the gold standard and has sensitivity of 60s to 70s, whereas PET - and bear in mind all the work done has been on instruments less sophisticated than are now available - 98% sensitivity in terms of diagnosis, a little less in specificity. Why? Because chronic inflammatory processes in granuloma will also take up glucose.

Early on, this was high because we had a more select population, but as the population increases, there is non-specificity. It's not that FDG is not specific; it is specific for glucose metabolism. Glucose metabolism is a good hallmark of neoplastic disease, since most tumors have increased anaerobic metabolism. It is also a hallmark of inflammation. The detection of recurrence is 98%, staging 82%. That's because it is size-limited. It ultimately depends upon what you are comparing it to, but it is more sensitive in CT or MRI to detect tumors in the mediastinum, lymph node involvement, and so on. CT and MRI diagnosis is based upon size, and now when you call disease in a normal size lymph node, unless there is something really bizarre about it, you saw in an earlier image that it will light up in red if there is tumor there or granuloma.

In the early application for colorectal, it was limited to detection of recurrence. You see a very high sensitivity, pretty good specificity, always better than CT. Lymphoma very good for staging. I have more to say about this in a moment - detection of recurrence - and the CT is good for staging, but FDG exceeds it. Even for staging and for recurrence, CT is useless, since it cannot tell the difference between nonviable tumor in a fibrotic tumor mass or recurrent tumor, and the specificity is quite high.

In melanoma, for staging we see that this number is a little less. It's interesting that there has been so little done with melanoma, but it is more specific than CT. And here we see in a small group, where they looked at this, it is all interesting. But in this cross containment area, what difference did it make? So here we see, in this population, in what percentage of the patients did the PET finally have an impact on management? And we see it's between 25 and 37%.

A few examples: On your left is a slide I showed earlier. This is, in fact, an old slide made on a GE scanner (not advertising - they gave it to me and I want to give them credit). This was the first patient done on this scanner. This was a young woman of about 30 who is the roommate of one of the engineers working on the project, and when they were looking for a volunteer and they were talking about it, I guess over dinner, she said she'd do it. She hadn't told anyone that, ten years before, she had had a melanoma removed and apparently no evidence of spread. It subsequently was found that she had some symptoms, but it was not known that she had disease.

We see the intense uptake in the brain that was typical because the brain uses glucose. This is an FDG scan in the patient. We see variable amounts of uptake, depending on what the patient has eaten, in the heart. I already discussed in disease but, in fact, if the patient has a high blood sugar, not high, but has just eaten carbohydrates, you will see cardiac activity. We are not seeing it in this patient;, in fact, we are seeing absent activity - seeing background glucose metabolism in the liver and soft tissues. This patient is resting. You see there is excretion of glucose in the kidneys. The one place that fluorodeoxyglucose differs from glucose is that the tubular epithelium does not recognize it and doesn't reabsorb it; so it is excreted and appears in the urine. This is bladder activity. This is an anterior coronal view, and these are deeper views. These spots are metastatic melanoma.

On your right is a slide from Memorial. One of the hallmarks in nuclear medicine is radioactive iodine imaging in thyroid cancer, and in the middle is short acting I-123 scan. This might be stomach activity. This is bowel activity. This really didn't characterize a tumor, although there might be something here. This is the FDG on this patient with thyroid carcinoma. This is the head. These are the shoulders, the mediastinal lymph nodes, paratracheal lymph nodes - intense activity. With information like this, where we can't get radioactive iodine in, what can we do? The management has changed and now surgeons will go in after these things often. This patient first had a high dose of radioactive iodine; so this might be done with a few hundred microcuries of 123. This was done with over 100 millicuries to treat the patient and, indeed, this tumor did take up. But we can see the routine diagnostic dose of FDG was much more sensitive and, in fact, detected disease that doesn't concentrate iodine very well, moving us toward alternate therapy for thyroid carcinoma like this.

This is the type of scan we were obtaining on our dual detector system. This is for our department - a patient referred with lymphoma, positive CT for a mass in the abdomen. This is coronal, sagittal, and transverse section through it with instrument like this. We modified staging, and we've used it for number of applications. We detected disease in lymph nodes as small as 7 mm with our instrument, and we published several manuscripts about it.

Some of the early applications have been thought not to replace CT in staging, although I think this would be one of the disease entities where the combined modality is used right away. Really, the value has been in detection of residual disease following therapy, because often the tumor mass may not completely resolve. The CT in this study in the Annals of Oncology, had an 86% sensitivity with no specificity and had 100% sensitivity a very good specificity. And it was viewed even in 1997 as the most useful technique for the evaluation of patients for residual disease, differentiating disease from scar tissue.

In lymphoma, there are several possibilities. One was initial staging. The other is evaluating for residual disease. There is still more. Is it possible, instead of having a patient go through - if you are familiar with the treatment for lymphoma - six, eight, or nine months of chemotherapy and then, or three months later, oops, we go into second round treatment after the marrow has been impacted, and so on? Is this useful to predict response? Let me show you what we have demonstrated in our department on our dual detector instrument.

This is a patient who, on diagnosis, has a baseline at an FDG scan, and we see this is renal excretion, extensive tumor activity, lymphoma, mediastinum, and a long superior mediastinum, probably supraclavicular. After the first cycle of CHOP [the chemotherapy regimen using cyclophosphamide, adriamycin, vincristine, and prednisolone], the scan has become normal, and at the end of therapy, of course, it's normal. This patient has been followed for more than 18 months so far and is still in remission.

Here is a patient with extensive disease. When evaluated after the last cycle, he was negative. As we work on this and begin to look at the first cycle, after the first cycle the patient still had residual tumor activity here. A lot of the tumor has been suppressed, the residual tumor. This patient relapsed within months, even though at the end of the therapy the scan had become normal. The way we interpreted this is that the chemotherapy can eventually club the tumor into submission reducing the glucose metabolism - this is eight months of CHOP therapy - but if it takes that much chemo to eliminate the glucose metabolism, these patients have a high frequency of relapse. The thinking now is that when we do this in a persistent disease for the first cycle, do we go on to the second line therapy right then? These patients have dural remissions, and we've reported on something like 40 of these patients who are still in follow-up. We reported it last June at the nuclear medicine meetings; it's being reported at the oncology meeting, with follow up, I think, a minimum of 18 months now and, in some cases, three years.

This is muscle activity. We've learned to recognize the patterns if the patient is uncomfortable and stressed. This is myocardium; it has to do with what they have eaten recently.

Will it show marrow?

It does show marrow. We have published an abstract demonstrating that we cannot simplistically use it to evaluate the presence of marrow disease because it will show marrow rebound after therapy, and nutrigen stimulates marrow. So we haven't figured out how to characterize disease. If there is uniform activity in the marrow, then we think it's hyperplastic marrow, following cessation of chemo. So, you know, we play games; when you walk in and you've seen it regularly, you see the whole marrow lighting up, you often may be right that it is a follow-up after therapy.

Of course, if the patient has osteoporosis and osteoarthritic problems with collapse and all, you will have a new regular pattern in the marrow, it can fool you. And we've shown then that it is unreliable to assume that a non-uniform simulation of the marrow means disease, but there are more surprises.

Just a few other applications: This is an example of unexpected mediastinal involvement or undetected involvement in a patient with non-small cell lung carcinoma. I've had many examples, but I didn't want to make it all oncology. It picks up adrenal metastases, unsuspected osseus metastases. This is a patient who had a lobectomy, a messed up X-ray and CT, and was being evaluated routinely. And this intense activity was found at the base, which was biopsied subsequently; this was a CT mystery, a recurrent tumor. By the way, we can adjust the glucose intensity to image the brain for a variety of disorders. This is a coronal section, a transverse section to show a patient who had a neck tumor involvement. It identifies disease in "normal" lymph nodes, a size below which CT is willing to call. This is pretreatment, which shows cessation of amino-acid accumulation, which is a marker for protein synthesis following treatment.

These images are not as good because the dosage you use in carbon-11 has a short half-life, so you have less information coming at you. There are a number of laboratories that have developed fluorine-18 labeled amino acids, and we expect that that will be available soon for glucose synthesis.

I want to move on and leave time for discussion.

Use of FDG in Parkinson's Disease

Just briefly let me show you some other things. These are old, probably ten years old from UCLA, on your left. It is a cross section of the brain, color-coded, increased activity in brighter colors in red. It is showing that when the patient - these are two different levels - when the patient is lying quietly with eyes closed, this is the pattern. When the patient is lying quietly with the eyes open, the accessory visual cortex lights up. This is eyes-open in a dimly lit room. The patient is looking at a complex scene. We see further glucose metabolism when the patient is lying there. I brought this one because it is interesting, and to show you that this can readily be quantified. And that it can detect the differences in quantitative terms between a patient with homonymous hemianopsia or eyes closed looking at wide light, looking at checkerboard pattern, one eye or two eyes, or a complex scene. This is percent increase in metabolic glucose utilization.

On your left is glucose, one slice through a brain comparing it to the same patient normal with fluorodopa to characterize the dopamine receptors and the basal ganglia. And here is the fluorodopa in a patient with Parkinson's disease, showing the reduction in the receptors in that disorder. This can be quantified and it is not approved clinically, but it has been very worthwhile in Parkinson's and other motor disorder research.

On your right is a pretty old slide given to me by the people at Hopkins. It shows another agent, I think it's haloperidol, labeled with fluorine-18, looking at a subset of dopamine receptors in a patient. It shows that in addition to the receptors in the basal ganglia, there are also a lot of distribution receptors, specific, nonspecific. The technique is evolved to do this in a basal condition and then to give a blocking dose of the non-radioactive drug and the second tracer to see what happens. In this patient, all of the activity was blocked by the drug, and I don't remember the details of what the amount of the drug was, but there was an excess of the drug. By picking doses in between, you can investigate a developed absolute displacement curve and characterize the affinity and the total receptor capacity. That is something I thought would be of interest to this audience.

FDG in Detecting and Following Inflammatory Disease

Now, a particular interest to this audience is a patient with temporal arteritis, reported from this institution in association with Memorial. A 75-year-old woman, I believe, who was well until a few months before, when she had the onset of low grade fever, weight loss, fatigue and generalized malaise. It was thought that she had a tumor somewhere and she was referred for an FDG image. This is a three-dimensional image, and that's why the quality looks so good, slices coronal sagittal, and you don't have to be a radiologist or a nuclear medicine physician to see this intense activity in a thoracic aorta - the arch and the descending aorta down to here. The major vessels, the subclavian vessels, in the report and even the left coronary artery, I was skeptical of that, but when the people sent me these images, I suppose this is what they had in mind. I have to admit I'm impressed, not because I didn't think there was activity there, just that this is a difficult thing to image - something of that caliber. So this is a patient reported from this institution, and Evan Leibowitz is co-author of the paper.

This is a patient who went on and demonstrated to have giant cell arteritis, I believe. And this shows the sagittal view, on the right, outlining the aorta quite vigorously, and following two weeks of prednisone therapy, we see reduction of the activity. So it addresses exactly the issues which Dr. Paget brought up. Can you detect disease? Can you use it to follow disease? This was not quantified, but it is showing you that we could quantify it.

I was also charged with telling you what else has been done. There have been some random reports like that. We indeed had a case last year. We didn't think it was worthwhile to report a single case, and it wasn't as beautifully imaged as that case, but I said to my friend Henry Young at Memorial "What else do you have? I'm supposed to talk to the folks at Special Surgery," and he sent this. This is a patient who had carcinoma of cervix with an onset of polyarthralgia and who was referred for imaging, and you can see her hands and wrists lit up. Her hip capsules lit up, and I think she is being viewed as rheumatoid arthritis.

Any active inflammatory process, we believe, will be imaged. When I saw the image a year ago with the aorta, I asked if the patient had active atherosclerosis going on. The reason I asked that was, when I was at Mount Sinai, we had become very interested in atherosclerosis and developed a conviction, even before it has become more widely viewed, that there was an inflammatory component in atherosclerosis. Dr. Schenker, who worked with me at Sinai then stayed for a few years and has now joined me at Cornell, he did a lot of that work. Meanwhile, the folks we left behind did this last year and reported it at the nuclear medicine meeting. A patient was referred for tumor, but they saw this intense activity, which corresponds to the arch and the descending aorta.

FDG in Atherosclerosis Imaging

Dr. Mahock, who is trained in internal medicine and nuclear medicine, was a resident in our program. He reported this abstract where he looked at 50 patients at Memorial and characterized the uptake by a technique called SUV and identified that there is a distribution and that age is just a weak correlation with increased uptake, and gender didn't correlate at all. In another group of patients, he found a weak correlation with obesity, with hyperlipidemia, no real correlation with diabetes or any of these other factors. But he did see the aorta in about 50% of the patients and likewise in a separate group of 50. So in atherosclerosis, something else is going on.

Dr. Schenker, before he came to Cornell, did this at Mount Sinai. He developed or used a rabbit model of atherosclerosis. These rabbits were fed a high cholesterol diets; all had hypercholesterolemia. This is a later view; so here is the aorta. When the animals also then had an aortic lesion, produced by excoriating the intima with balloon, this is the kind of intense activity they developed. So both are hypercholesterolemic, but the balloon injury produced an inflammatory response with increased uptake. This is of a glucose of FDG.

This is comparing the normal rabbit aorta to the excoriated aorta in the hyperchloremic animal - ex vivo images. And here is some quantitative information where he compared the percent of the injected dose in the aorta, in the green in the normal rabbit and the hypercholesterolemic rabbit. This is using radio-labeled LDL. This is using radio-labeled oxidized LDL. We see marked increase in the hypercholesterolemic animal.

Here he looks at what the FDG uptake was in the hypercholesterolemic animal rabbit aorta, and it's markedly increased. He went on to do histologic characterization with pathologists quantifying the FDG in slices, aortic segments, and comparing it to, (I don't have the details as to methodology) and scoring it for macrophage density. It is our belief that the increased glucose metabolism and the active atherosclerotic aorta or the inflammatory aorta is due to the macrophage activity.

FDG to Evaluate Loosening Prostheses vs. Infection

Now we can stop here, or I can briefly show you some work which I showed last year from Chris Palestro, who is now at Long Island Jewish. I'll tell you briefly what it is. About loosening prosthesis versus infection. So in this patient you see the white cell is positive. We make sure it's not marrow, and the FDG is positive. This prosthesis was removed, and indeed was infected, with positive cultures and so on. He has been the only one with the courage to say, however, this is nonspecific. Here is someone with a negative white cell, positive bone scan. Loosening or infected? The white cell says it's not infected; the FDG shows increased uptake. This prosthesis was removed, and it was culture negative. There was a proliferative response, but it wasn't polys. When we label the white cells, we are labeling the polys. When we are imaging FDG, we are imaging whatever is there, and indeed there are a lot of macrophages. And we know that there is a sterile inflammatory response to loosened prostheses. So he and I and others believe that it is not specific for infection; it's demonstrating inflammation and macrophage response.

So I will stop there and take questions. Thank you. We have a few requirements. We need lights and we need a microphone.

Question and Answer Session

I gather from your statement that you initially thought that aorta was atherosclerotic, but you can't necessarily tell aortitis from atherosclerosis.

Exactly. Well, we can tell by looking at the image. We haven't done much characterization, you know, quantification, and you will be in a position to do that. We are in a holding action with the instrument that we have now. When we have the higher resolution instrument at the end of the year, we can begin to do it. It will be easier to quantify or compare it to background of some other organ, a referenced organ, and it may be a difference. I would think, as you know, that atherosclerosis-- that many people think it's a macrophage disease.

Most of our patients with temporal arteritis are elderly and have atherosclerosis in their aorta plus or minus arteritis.

I hope you will help us solve that problem.

I have two questions. Is this reliable in insulin dependent diabetes? And the second questions is: what is the status of central nervous system lupus?

Okay two questions, one: is it reliable in insulin dependent diabetes? It's reliable depending upon what the blood sugar is, but also the insulin status. In a hyperglycemic patient, which is greater than 200 or 250, you are having mass action, competition of the glucose with the FDG. A mistake, however, is to give the insulin right then and lower it. Because when you do that, you will send all the FDG into the muscles. We have slides of that. We have routinely done finger sticks. I do it where they are very busy, where they are trying to do 15 patients a day. They don't do it if the patient doesn't know they have diabetes. They do it, and most of the time they haven't been burned. We've had an asymptomatic man who had a blood sugar of 300 and didn't know he had diabetes until then.

What we do is give insulin and wait until the blood sugar stops dropping. So once it's stable and the glucose has been driven into the muscles of the liver and stabilized, then we use the FDG as a marker. So it's more work. You can detect tumors. Now, what the influence would be on something like arteritis remains to be seen. I think you'll be able to detect a process if you've lowered the blood sugar to below 200 and it's an equilibrium.

The second question was central nervous system lupus. We had a long conversation about that. No one's looked at it, and we are eager to look at it with you.

There are two aspects to central nervous system lupus. One is the inflammatory component. I showed you that just thinking about certain things or looking at a complex scene will change your glucose metabolism in parts of the brain. So we would have to sort that out. In other words, if there was arteritis last year and many infarcts, you'll have decreased perfusion. Glucose is the universal substrate for the brain So you would have to differentiate between cortical utilization of glucose and cerebral arteriolar inflammation.

Now the other thing that I would hope to impart is that although glucose is available to us readily, and very powerful, it's not the only positron emitting radionuclide. And it really falls on us to use our imaginations to identify which ligands or other substrates would help unravel the question and then develop a method to synthesize and characterize normal versus abnormal. So we can think of all kinds of things. As far as I know, prostaglandins have not been looked at. But in the imagination, we don't have to label prostaglandin, or we could do a prostaglandin and look at the response to a drug which might inhibit prostaglandin synthesis, a COX-2 inhibitor, for example. Dr. Meiselas has a question, from which we always learn.

Meiselas: Stanley, stop it. Anyway, when you talked about the movement of the eyes producing different hues of glucose, what about when somebody is thinking? How do you manage that? How are you sure that somebody is not thinking "Hey, why don't you leave me alone" or something like that?

I know what you are thinking, and some of this answer is classified. I was very tempted to show more of that brain stuff, because it's very interesting. It's great cocktail party stuff, but it wouldn't allow us any time for questions. There are different regions, and if you saw yesterday's Science Times, the work actually began in nuclear first with FDG and then moved on to oxygen-15 water. And Dr. Silbersweig at Cornell and Emily Stern were working together and demonstrated all kinds of things - patterns of audio and visual hallucinations in schizophrenics, and comparing the reports on various drugs and all that, looking for objective evidence of hallucinations and what is going on. You can tell by locations. Some of that work has drifted over to MRI, where you can see small changes in perfusion. That is what you are seeing with oxygen-15 water, and it's just easier to keep re-doing the MRI without any radiation. You just have the magnetic field problem and the flying oxygen tanks and things like that. It's one of the reasons investigators are interested in a question like yours. They are so eager to switch to oxygen-15 water. So they want to do what is called trial and retrial, test and retest.

So you take a baseline image, and you repeat it five times or six times and you see what background change is going on. I have a slide showing auditory stimulation. There are slides of a musical tone or noise. In yesterday's Times, they tried to look at what parts of the brain people use to solve what they call dilemmas. It depended upon whether it was a personal dilemma, interaction with a person, or an impersonal interaction. So what happens when you make a decision? The end result is the same number of people survived and the same number of people died. If all you did was press a button, people use one part of the brain, but if you have to push someone off the ledge vs. saving five other people, you use another part of the brain. It was very interesting. It's a whole area. I think you have a future in cognitive psychology.

I have a comment and perhaps there is a question involved with the comment. One doesn't have to do a PET in order to make a diagnosis of atherosclerosis. You've got Imatron, a new imaging tool that has been all over the world, and it demonstrates calcification and supposedly evidence of atherosclerosis in patients who are routine patients. Now what do you think about that?

I'm glad you asked that because that is a lob right over the plate. Calcification is usually present when you have atherosclerosis. So it's a marker of it. But having calcium doesn't mean that you have atherosclerosis. Having atherosclerosis means you probably have calcium.

Are there any studies showing if the patient has a certain degree of interstitial lung disease, whether they have active alveolitis or not?

I haven't seen images of alveolitis, but I would expect it to be positive. It would be a diffuse inflammatory process and, in sarcoid, certain nodal involvement, sarcoid is widely positive. It cannot be differentiated from lymphoma by pattern, although like everything else, some nuclear medicine people have some clinical judgment.

If you can tell me what's unique, to return to connective tissue inflammation, what is unique about that, then we can develop a radio label to look at it. One of the things that is relatively unique is macrophage infiltration. Macrophages use glucose. So glucose won't give us specificity simply by uptake, although by that pattern, next time I see an aorta that intense and that homogenous, I would say giant cell arteritis before I would say atherosclerosis. If it's more patchy, I'd say atherosclerosis. That is adding a clinical hunch to an image.

If there is nothing to stop us from identifying a peptide or an antibody that recognizes a unique epitope or receptor on a macrophage and radio labeling it, now we don't need PET for that. But PET gives us great sensitivity, and in addition to being inherently tomographic, it's inherently quantitative, whereas other nuclear medicine is more difficult to quantify.

The other advantage of PET is that we are throwing two million dollars at the problem. So the images that we get are very good. We have exquisite computer handling of data, instrument displays and a lot of convenient image processing and then, when we have it added to the CT, that will be great. There is a lot of software that comes with it, which we don't have enough time to review, but will be very useful with FDG. I would look to things like diseases of macrophage function. I would look to characterize receptors or epitopes.

Steven A. Paget, MD: Just really a comment. A lot of the questions have dealt with diagnostic use of this. But I really want to emphasize what I think are exciting potential research applications to define pathophysiology. And specifically to the question that Bento asked, the plans that we have are to try to dissect out, at the receptor level, mechanisms for cognitive dysfunction in lupus patients, in those patients who do not have anatomically recognizable disease. The way that Stanley has been able to create ligands that will identify what's happening to specific areas of the brain, I think will advance our thinking tremendously.

That is a very good summary remark.


Related Professional Content