Radiology #200308003
Dr. Michael Graham
PET Imaging of Solitary Pulmonary Nodules
Abstract:
This project is an un-funded continuation of Specific Aim 4 of an NIH-funded study on PET imaging in esophageal carcinoma (CA74959). We have closed subject accrual into the esophageal cancer FDG PET study and are now analyzing the data. The funded duration of the original grant was through June 2002. Funding was extended for one year through a no-cost extension. We now have a 2nd no-cost extension with residual funds to finish the data analysis of the esophageal cancer study. The original grant proposal included specific aim 4, which was to extend the imaging methodology used in the esophageal cancer study to the diagnosis of indeterminate pulmonary nodules. This IRB application is to obtain preliminary data for SA4. The subjects will be imaged as the first PET study of the day, which will cause minimal disruption of the PET schedule and will have no impact on the number of clinical studies we can do.
Quantitative measures of uptake of FDG into pulmonary nodules have been used for several years as a method to objectively determine the probability that a nodule is malignant. Our group and others have reported sensitivities of his approach of around 90% with specificities ranging from 60 to 90%. Our specificity in a recent study was 65%. We ascribe this relatively low specificity to the high incidence of histoplasmosis in the Midwest. There are three approaches that we are proposing to improve the accuracy of the method, particularly to improve specificity. These are: use more sophisticated data analysis methods, image at a later time after injection of FDG, and use the rate of change of uptake as a predictor of nodule status. Surprisingly, remarkably few studies of this type have been done. The usual measure of uptake is the Standardized Uptake Value (SUV), and most sites doing PET imaging use this. Several groups have looked at SUV correction schemes but there is virtually no work-using tumor to normal. Several groups have looked at SUV correction schemes but there is virtually no work using tumor to normal tissue ratios, simplified kinetic analysis, or Patlak analysis. The latter two methods are ones that require a blood sample and are more accurate measures of glucose metabolic rate than SUV. We retrospectively studied 74 subjects who had FDG-PET studies for pulmonary nodules and found tumor to normal tissue ratio was slightly, but not statistically significantly, better than SUV in determining the status of the nodules.
It is quite well known that the SUV of tumors rises with time after injection. It seems reasonable that it should be easier to diagnose tumors if imaging is done later. Some groups have advocated imaging as late as three hours after injection, although no one has looked to see if it really is more accurate. About 18 months ago we changed our clinical protocol to conduct PET images at 90 minutes post injection, having previously done the images at 60 minutes. We retrospectively compared the results from 42 subjects imaged at 60 minutes with 58 at 90 minutes and found the sensitivity remained constant at 84% at the specificity fell from 75% at 60 minutes to 65% at 90 minutes. This suggests that later many not be better, however it is very important to repeat this in a paired study, where the results should be very statistically significant.
A group at the University of Pennsylvania studied 35 pulmonary nodule subjects (19 positive, 16 negative) at 70 and 120 minutes after injection. They showed standard PET scanning (single time point) with a threshold SUV of 2.5 (at time point 1) reached a sensitivity of 80% and a specificity of 94%, while dual time point scanning with a threshold value of 10%increase between scan 1 and scan 2 reached a sensitivity of 100% with a specificity of 89%. This work has not been replicated by any other group.
We are proposing to study 100 subjects with indeterminate pulmonary nodules. The N is based on a MedCalc power calculation that, to detect a change in sensitivity from 75% to 90%, with alpha = 0.1, beta = 0.2, we need 77 subjects (MedCalc Ver 4.31). In past studies at the University of Iowa approximately on third of the nodules were benign. The subjects will be imaged at 60, 90, and 120 minutes after injection with FDG. A blood samples will be taken at the time of each PET image and counted for radioactivity. From the images and the blood sample counts we will be able to calculate 5 different measures of uptake at each time point, an overall measure of uptake (Patlak), and the differences between all individual measures. The 5 measures are: SUV, simplified kinetic analysis, and nodule to normal tissue ratios using contralateral lung, liver, and mediastinum as the normal tissue. A total of 31 parameters including differences, will be calculated for each subject. True diagnosis will be determined by pathology from biopsy or resection, or follow-up of greater than one year. Once the status of the nodules is determined, the different measures will be compared for diagnostic accuracy using receiver operating characteristic (ROC) curves.