Room: Karl Dean Ballroom C
Purpose: [F-18]-FDG PET/CT, CT, and MRI images are used clinically to determine progression of advanced lung cancer (ALC) patients. Previous work has shown that FDG PET/CT scans add value to response assessment, but heterogeneity of response has not been evaluated. This study examines the response heterogeneity seen on FDG PET/CT and assesses its agreement with evidence of progression seen on CT and/or MRI.
Methods: Eleven ALC patients treated with systemic therapies received FDG PET/CT scans at baseline and within 41 (median 1) days of progression (CT, MRI). An experienced nuclear medicine physician identified PET lesions and verified segmentations produced by an automatic tool. Relative change (%Î”) in SUVmax, SUVmean, SUVtotal, and PET volume were used to classify and measure lesion- and patient-level response. Lesions and patients were classified as completely metabolically responding (CMR, disappearing), partially responding (PMR), stable (SMR), progressing (PMD), or new (NMD) based on published repeatability limits of agreement. Patient-level metrics were determined using lesion segmentations. Fleissâ€™ kappa was used to assess agreement between response categorizations.
Results: Patients demonstrated heterogeneous FDG response: 7/11 patients had both CMR/PMR and PMD/NMD lesions, 4/11 had both SMD and CMR/PMR or PMD/NMD lesions when response was categorized by %Î”SUVtotal. Patient-level PET/CT response determined by %Î”SUVtotal had the greatest agreement with radiographic progression: 4/11 were classified as PMD, 2/11 as SMD, and 5/11 as PMR. The discordant patients either had new lesions (identifiable on FDG PET) or progressing brain metastases (not identifiable on FDG PET). Patient-level response was more discordant when assessed by the other metrics. Response of patients and lesions had fair agreement when classified by different metrics (Îºpatient=0.22 and Îºlesion=0.34).
Conclusion: Response heterogeneity was observed in FDG PET/CT scans of eleven ALC patients. Further work is needed to combine quantitative response information from other imaging modalities (MRI, CT).
Funding Support, Disclosures, and Conflict of Interest: Work funded in part by the University of Wisconsin Carbone Cancer Center Support Grant P30 CA014520 and the Wisconsin Oncology Network of Imaging eXcellence.