Virtual Library
Start Your Search
J. Fledelius
Author of
-
+
O23 - Imaging and Screening (ID 125)
- Event: WCLC 2013
- Type: Oral Abstract Session
- Track: Imaging, Staging & Screening
- Presentations: 1
- Moderators:J.R. Jett, H.M. Marshall
- Coordinates: 10/29/2013, 16:15 - 17:45, Bayside 201 - 203, Level 2
-
+
O23.04 - Improved interobserver agreement with PERCIST 1.0, compared to a qualitative method for early response evaluation using FDG-PET/CT in NSCLC (ID 1195)
16:45 - 16:55 | Author(s): J. Fledelius
- Abstract
- Presentation
Background
During the past decade many studies have used FDG-PET/CT for response evaluation to therapy, both in NSCLC and several other cancer forms. A preferential method for evaluation has not been established as of yet. Two main approaches tend to single out: A visually based model as described by Mc Manus et al in 2003 and semi-quantitative approaches, like the recently proposed PERCIST 1.0 2009, by R. Wahl et al. Few studies have evaluated the interobserver variability when using sequential PET/CT scans for response evaluation, and comparison of qualitative- and semi quantitative approaches are also scarce. The aim of this study is to determine which method will provide the more robust evaluation of response when using FDG-PET/CT, the qualitative approach or the SUV based semi quantitative approach prior to the introduction of routine early response evaluation in NSCLC.Methods
FDG-PET/CT scans at baseline and after 2 cycles of chemotherapy from 35 patients with locally advanced NSCLC were analysed by 8 different readers using two different methods: PERCIST 1.0 and the qualitative McManus approach. Both methods result in allocating patients into one of four response categories. Observers were given short written presentations outlying the criteria for evaluation by the two methods. The observers represent a wide range in experience with PET evaluation, only half had experience in response evaluation in NSCLC, but most were experienced in similar evaluation in lymphoma patients.Results
When using PERCIST 1.0, the agreement between observers in determining the percentage change in SULpeak was “almost perfect” with ICC=0.959 similar ICC values were found looking at SUL peak at baseline and follow up scans. There was strong agreement amongst readers allocating patients to the different response categories with Fleiss kappa of 0.761 (0.714-0.808). In 22 of the 35 patients there was complete agreement. When using the qualitative method (A.M. McManus), agreement was lower, down to moderate agreement, with Fleiss kappa of 0.596 (0.554-0.639). And complete agreement was observed in only 10 of the 25 patients. Using chi squared the difference is statistically significant (p < 0.005). No difference was found between experienced and non-experienced observers.Conclusion
In spite of a wide range of experience among 8 readers receiving minimal introduction to the two methods they were to use, we found rather high kappa values, for both methods compared to its nearest competitor: Size change in CT images, known to be very observer dependent. The more objective, semi-quantitative method showed substantially higher agreement than the more subjective method. We suspect that a more detailed introduction into the methods would have improved the kappa values even further, but believe that our method is more likely to provide an introduction similar to the one you receive when introduced as a new physician at the department. Perhaps then the agreement reflects the long-term agreement.Only Members that have purchased this event or have registered via an access code will be able to view this content. To view this presentation, please login, select "Add to Cart" and proceed to checkout. If you would like to become a member of IASLC, please click here.
