IMAGING SECTION II - CHARACTERIZATION OF SMALL PULMONARY NODULES WITH PET AND CONTRASTED CT SCANNING, A. PET AND SPECT IMAGING by Barry Siegel, MD

SLIDES & TRANSCRIPTS
Tuedsay, June 19

IMAGING SECTION II - CHARACTERIZATION OF SMALL PULMONARY NODULES WITH PET AND CONTRASTED CT SCANNING, A. PET AND SPECT IMAGING

Barry Siegel, MD

Slide 1: Nuclear Medicine Approaches

DR. SIEGEL: Well, I'd like to say that it would be possible with nuclear medicine to say as many elegant things about what we are going to do with these tiny nodules as you just heard about with computer-assisted CT. But unfortunately that really is not the case, at least where technology stands at the moment.
So what I'm going to do in the next few minutes is just tell you a little bit about what we do know regarding the use of nuclear medicine approaches for characterizing lung nodules, and then pose a few questions about things we need to study as we get into the screening detected problem.

Slide 2: Conventional Nuclear Medicine Methods

Conventional nuclear medicine methods have been applied for characterizing lung nodules for a number of years. There are a couple of very old tracers, gallium and thallium, which basically are of no real interest anymore, because their imaging characteristics just don't lend themselves to dealing with very small lesions.
There are a variety of technetium-labeled tracers, with reasonable results, but none of these have caught on to any degree. A number of years ago there was quite a bit of interest in using monoclonal antibody imaging, and some results showed reasonable sensitivity and specificity. Again, this is an area that at the moment, has faded, because metabolic imaging has moved into the forefront. I think monoclonal antibodies is an area, and other biospecific markers are an area that will certainly need to be investigated to try to improve specificity, as I will comment near the end.

Slide 3: Somatostatin Receptor Imaging

An area that has moved into clinical practice is a biospecific approach known as somatostatin receptor imaging. There has been work done in the past with the Indium-111 labeled octreotide component particularly in small cell lung cancer, but also in non-small cell lung cancer.

Slide 4: Multicenter Study

And the most recent interest has been in this technetium-labeled synthetic decapeptide, the technetium-99M depreotide, or Neotech as it is known commercially. As you can see, it's a synthetic somatostatin analog. It binds to somatostatin receptor subtypes 2, 3, and 5. It has been shown to have good uptake in non-small cell lung cancer. But the mechanism of uptake is really not clear at the present time. There is expression of somatostatin receptor type 2 in a very small fraction of non-small cell lung cancer, and the expression may be related more to peritumoral inflammatory cells, some endocrine cells present in the tumors, or one study suggested that these receptors actually were in peritumoral veins.

Slide 5: SPECT Image

There has really been only one large published study with this radiopharmaceutical that looked at 114 indeterminate lung nodules. The smallest ranged down to 0.8 centimeter, and that was a benign lesion. Imaging was done both with planar scintigraphy, as well as SPECT. All the lesions were confirmed by a biopsy. The sensitivity was 97% with false negative results in two primary adenocarcinomas, and one metastatic adenocarcinoma, while all of these lesions were under 2 centimeters. I think the smallest was 1.1 centimeter. Specificity has not been quite as good, however, with a specificity of only 73%. And the false positives were largely related to granulomas, which has been the bane of most of the nuclear medicine approaches, and one hamartoma. The explanation for that is not at all clear. Just a couple of quick examples from the published multicenter study. Here is a fairly good sized adenocarcinoma, which is easily seen on the SPECT image.

Slide 6: SPECT Image

Here is a fairly large neurofibroma with no abnormal uptake seen on this SPECT image.

Slide 7: PET

Far more interest in recent years has focused on PET. PET offers the advantage by comparison to conventional nuclear medicine techniques of coincidence detection of annihilation radiation, thereby getting rid of the limitations imposed by the collimator. A higher intrinsic photon detection efficiency, and the ability with the right kind of instrument design, to achieve higher spatial resolution than can be achieved with conventional nuclear medicine imaging instruments for the most part, ultimately limited by the range of the positrons themselves. And that becomes a problem when you move away from things like fluorine-18.

Slide 8: Glucose Metabolism

Because of the coincidence detection, it is possible to do very precise attenuation correction in PET. And that means that PET results can be more precisely quantitated than those done with conventional nuclear medicine imaging, planar imaging, and SPECT.

Slide 9: FDG Uptake

For all practical purposes at the present time, clinical PET is the assessment of glucose metabolism with F18 labeled fluorodioxyglucose. In many ways that's a mistake, and hopefully we won't end up there over the long-term, but for a variety of logistical problems and FDA-related problems, this is where clinical PET is at the moment.

Slide 10: Evaluation of Tumor Uptake

I won't waste time really reviewing the mechanism of FDG uptake, but basically it's a glucose analog that gets into cells, and in malignant cells, brain and myocardium in particular, once it gets in there and is phosphorylated it stays there. Its integrated uptake over time is effectively related to the glucose metabolic rate of the tissue.

Slide 11: SUV

There are several different ways to evaluate FDG uptake in tumors. Clinically, for the most part this is done by just qualitative visual assessment. There are semi-quantitative methods, and much of the pulmonary nodule literature has focused on the standardized uptake value, which I will mention in a moment. Then there is more complicated quantitative methods that pay attention to the input function, and these have rarely been used in the setting of lung nodule assessment, but there may be some role for these, as several investigators have shown that there are differences in the behavior of malignant lesions with respect to the uptake and retention of FDG when compared to inflammatory lesions. And this is one research area that will need to be explored. The SUV is basically just an expression of the concentration of the tracer and the nodule, divided by the average concentration of the tracer and the body, assuming there was no excretion. So an SUV of 1 indicates that the voxel of interest is equal to the average concentration. A number greater than 1 means that there is local concentration of activity there. Of the magic number of 2.5 has appeared in a lot of the pulmonary nodule literature. It's an entirely arbitrary number, and really has not been as well validated as it needs to be.

Slide 12: Examples

A couple of examples from one of the early papers done by the group at Duke University, showing above a very brightly seen nodule in lingula, and below a right lower lob nodule, with barely perceptible FDG uptake of the upper one being a cancer, and the lower one being a benign inflammatory scar-type lesion.

Slide 13: FDG PET

There has recently been a meta-analysis of the published literature of FDG PET in assessment of pulmonary nodules and mass lesions. The group at Stanford looked at 40 published studies involving about 1,500 focal pulmonary lesions. It is not surprising, given the way most of these studies were funded, which is off the backs of individual clinical PET facilities, that not all of the studies met -- in fact, none of the studies met all of the criteria defined by the analysts for an adequate quality study.

They did find that the maximum joint sensitivity and specificity was in the 90s. And the operating point in current practice for all lesions shows a sensitivity of about 97%, with a specificity of about 78%. Notably, they found no difference in whether one did this qualitatively or quantitatively, but very importantly for the discussion at hand today, is there is extremely limited data on lesions of less than a centimeter. In fact, they comment that they are able to get data out of only 8 lesions in this 1,500, where they can precisely find what the results for lesions less than a centimeter.

Slide 14: Summary ROC Curves

These are the summary ROC curves from this paper showing all lesions on the left, and the nodules defined as lesions looking like nodules of less than 4 centimeters in diameter on the right. Showing the joint sensitivity and specificity, and then the presumed operating points namely being the sensitivity at the median specificity in these studies. So already I have mentioned the 97% and the 78%. For nodules it's 94% and 84%. So sensitivity is reasonably good, specificity is a problem.

Slide 15: Limitations of FDG PET

There are obviously limitations of FDG PET for lung nodule characterization. First, the false negative results. And that seems to be largely a problem of small lesions, and the problem of partial volume averaging. There are some hypometabolic neoplasms. I think in practice this has been misinterpreted often to assume that such lesions are almost always negative. And in fact, that's not the case. It's just that they are occasionally negative, and on average they have lower SUVs than do the garden variety of carcinomas. The problems are with mucinous carcinomas if you will. This is a form of microscopic partial volume averaging. A cell taking up FDG, surrounded by a larger volume of mucin, and then another cell over there. So their signals get averaged over that mucin. Bronchoalveolar carcinomas have been notoriously bad actors, as are some of the low grade neuroendocrine tumors. A problem that will have to be dealt with when we get out into the large population is the diabetic patient. Hyperglycemia and hyperinsulinemia both are really very detrimental to doing a good job with FDG imaging, with markedly lower tumor uptake. And then we have the problem of false positive results, and far and away this is related to inflammatory lesions, and far and away that is related to active, sometimes very low activity, but active granulomatous disease.

Slide 16: Tumor Biology and FDG

We do learn a little something about tumor biology with FDG imaging. One study from Duke has shown a reasonably good relationship between the doubling time of lesions and the FDG uptake. This is not just malignant tumors. They also showed that there was a relationship with benign lesions as well. SUV is also related in a couple of published studies to prognosis.

Slide 17: Problems with Nuclear Medicine Approaches

The problems we've got to deal with are this uncertain sensitivity for small lesions. We just don't know. No one has looked at this specifically. We also need to make our scanners perform better. And then we have got this problem of inadequate specificity. We need tracers that localize in tumors, but not in inflammatory lesions.

Slide 18: PET Scanner Performance

To get a handle on scanner performance, there is a lot of technical innovation finally occurring now that PET has become something widespread in clinical practice. We are looking at several new scintillation crystals, particularly LSO and GSO, and the combined PET CT scanner, which gives us information about better attenuation correction, and the ability to fuse the anatomic and metabolic data.

Slide 19: ECAT Accel

Here, an example from some images from an LSO scanner. And let me just tell you that you can get good image quality in times that are substantially shorter than with current standards. But if you are looking for small lesions, you would take longer scanning times to get better quality images, which is how you try to attack the small lesions.

Slide 20: Inline Combined PET-CT Scanner

This is an example of a clinical study with an inline combined PET-CT scanner showing the ability to register those two data sets, which will be I think is very important for the small nodules. It allows us to apply recovery coefficients of lesion size from the CT to the PET data. It allows us to find the lesion on CT that we need to evaluate on PET, and what we have to address, the respiratory registration problem.

Slide 21: Tumor Metabolism Assessment with PET

FDG is clearly not the whole answer. These are just metabolic tracers, not even beginning to get into the list of things that can be looked at for various receptors. And there is really a lot of opportunity to evaluate some of these things.
My time is up, so I'll be quiet. Thank you.