2413: Multiobserver Variabilty in Abnormal Lymph Node Identification With FDG-PET and Contrast Enhanced CT in Head and Neck Cancer: Qualitative Analysis and Potential Clinical Significance

Abstract

Purpose/Objective(s)The treatment of head and neck squamous cell cancer (HNSCC) with conformal radiation requires precise identification and localization of the primary tumour and gross lymph node (LN) metastasis. PETCT is more sensitive than conventional imaging in staging the neck in HNSCC and may contribute to target delineation of LN for radiotherapy. The present study describes the differences in identification of LN considered to be involved in node-positive HNSCC using contrast enhanced CT (CECT) compared to PETCT.Materials/Methods10 patients with N2 or N3 HNSCC underwent PET-CT simulation, immobilized in treatment position to obtain CECT and FDG-PETCT images. 8 specialists (6 Radiation Oncologists and 2 neuroradiologists) independently contoured abnormal LN on each image set, aided by anonymised clinical and radiological patient information following a schedule to minimize memory bias. The specialists were asked to contour only those LN considered to be involved on the basis of imaging characteristics and/or relationship to the primary tumour. In the absence of a pathologic correlate, observer consensus was used as a surrogate for lymph node involvement as follows: LN structures identified by at least 6 of 8 observers on either modality were considered ‘positive’, identification by only 4 or 5 of 8 observers was considered ‘equivocal’ and LN identified by fewer than 4 observers, were considered ‘negative’. Any disparity between imaging modalities was reported. A significant disparity was considered to exist when a LN structure was deemed positive on one modality and negative on the other.ResultsDisparities between LN identified on CECT compared with PETCT were observed in 6 of 10 cases. In 3 of these cases the disparities were considered significant with PETCT identifying LN not deemed to be involved on CECT (6 LN in 3 cases). In one of these 3 cases CECT identified as positive, a LN that was considered negative on PETCT. In 2 of the remaining cases with disparities, LN were considered positive on CECT and equivocal on PETCT or equivocal on CECT and negative on PETCT. In one case a LN considered equivocal on PETCT was deemed negative on CECT.ConclusionsDiscrepancies exist between CECT and PETCT in the identification of LN thought to be involved. PETCT identified additional LN not appreciated to be abnormal on CECT in 3 of 10 cases. Significantly, in one case the CECT was positive while the PETCT was negative. Less significant but similar differences were observed in two other cases, suggesting the importance of considering complementary information from both modalities when determining LN volumes for radiotherapy treatment. Purpose/Objective(s)The treatment of head and neck squamous cell cancer (HNSCC) with conformal radiation requires precise identification and localization of the primary tumour and gross lymph node (LN) metastasis. PETCT is more sensitive than conventional imaging in staging the neck in HNSCC and may contribute to target delineation of LN for radiotherapy. The present study describes the differences in identification of LN considered to be involved in node-positive HNSCC using contrast enhanced CT (CECT) compared to PETCT. The treatment of head and neck squamous cell cancer (HNSCC) with conformal radiation requires precise identification and localization of the primary tumour and gross lymph node (LN) metastasis. PETCT is more sensitive than conventional imaging in staging the neck in HNSCC and may contribute to target delineation of LN for radiotherapy. The present study describes the differences in identification of LN considered to be involved in node-positive HNSCC using contrast enhanced CT (CECT) compared to PETCT. Materials/Methods10 patients with N2 or N3 HNSCC underwent PET-CT simulation, immobilized in treatment position to obtain CECT and FDG-PETCT images. 8 specialists (6 Radiation Oncologists and 2 neuroradiologists) independently contoured abnormal LN on each image set, aided by anonymised clinical and radiological patient information following a schedule to minimize memory bias. The specialists were asked to contour only those LN considered to be involved on the basis of imaging characteristics and/or relationship to the primary tumour. In the absence of a pathologic correlate, observer consensus was used as a surrogate for lymph node involvement as follows: LN structures identified by at least 6 of 8 observers on either modality were considered ‘positive’, identification by only 4 or 5 of 8 observers was considered ‘equivocal’ and LN identified by fewer than 4 observers, were considered ‘negative’. Any disparity between imaging modalities was reported. A significant disparity was considered to exist when a LN structure was deemed positive on one modality and negative on the other. 10 patients with N2 or N3 HNSCC underwent PET-CT simulation, immobilized in treatment position to obtain CECT and FDG-PETCT images. 8 specialists (6 Radiation Oncologists and 2 neuroradiologists) independently contoured abnormal LN on each image set, aided by anonymised clinical and radiological patient information following a schedule to minimize memory bias. The specialists were asked to contour only those LN considered to be involved on the basis of imaging characteristics and/or relationship to the primary tumour. In the absence of a pathologic correlate, observer consensus was used as a surrogate for lymph node involvement as follows: LN structures identified by at least 6 of 8 observers on either modality were considered ‘positive’, identification by only 4 or 5 of 8 observers was considered ‘equivocal’ and LN identified by fewer than 4 observers, were considered ‘negative’. Any disparity between imaging modalities was reported. A significant disparity was considered to exist when a LN structure was deemed positive on one modality and negative on the other. ResultsDisparities between LN identified on CECT compared with PETCT were observed in 6 of 10 cases. In 3 of these cases the disparities were considered significant with PETCT identifying LN not deemed to be involved on CECT (6 LN in 3 cases). In one of these 3 cases CECT identified as positive, a LN that was considered negative on PETCT. In 2 of the remaining cases with disparities, LN were considered positive on CECT and equivocal on PETCT or equivocal on CECT and negative on PETCT. In one case a LN considered equivocal on PETCT was deemed negative on CECT. Disparities between LN identified on CECT compared with PETCT were observed in 6 of 10 cases. In 3 of these cases the disparities were considered significant with PETCT identifying LN not deemed to be involved on CECT (6 LN in 3 cases). In one of these 3 cases CECT identified as positive, a LN that was considered negative on PETCT. In 2 of the remaining cases with disparities, LN were considered positive on CECT and equivocal on PETCT or equivocal on CECT and negative on PETCT. In one case a LN considered equivocal on PETCT was deemed negative on CECT. ConclusionsDiscrepancies exist between CECT and PETCT in the identification of LN thought to be involved. PETCT identified additional LN not appreciated to be abnormal on CECT in 3 of 10 cases. Significantly, in one case the CECT was positive while the PETCT was negative. Less significant but similar differences were observed in two other cases, suggesting the importance of considering complementary information from both modalities when determining LN volumes for radiotherapy treatment. Discrepancies exist between CECT and PETCT in the identification of LN thought to be involved. PETCT identified additional LN not appreciated to be abnormal on CECT in 3 of 10 cases. Significantly, in one case the CECT was positive while the PETCT was negative. Less significant but similar differences were observed in two other cases, suggesting the importance of considering complementary information from both modalities when determining LN volumes for radiotherapy treatment.