Abstract
Although, the 8th edition TNM (TNM8) significantly improved risk stratification of HPV-positive oropharyngeal carcinoma (HPV+ OPC) compared to the 7th edition TNM (TNM7) which was agnostic to tumor HPV status, undesirable heterogeneity still exists within each N category of the TNM8 classification, especially in cN1 disease. Compelling evidence suggests that image-identified extranodal extension (iENE) is prognostic in HPV+ OPC. The International Collaboration of Oropharyngeal cancer Network evaluated its role in refining the TNM8 cN classification in HPV+ OPC (ICON-N-HPV+).
Curative-intent HPV+ OPC from 4 institutions in Canada, USA, and the Netherlands were included. cN+ patients were randomly allocated to Training (60%) and Validation (40%) datasets. All cases were staged according to TNM8. Pre-treatment (within 8 weeks of treatment commencement) CT/MRI were reviewed by radiologists for presence/absence of iENE (iENE+/iENE–) and other nodal features, including total lymph node (LN) number, laterality of neck nodes, presence of retropharyngeal LN [RPLN]) using a priori definition. Multivariable analysis (MVA) employing a step-wise approach evaluated the prognostic value of iENE and other nodal features for overall survival (OS) in cN+ patients in both Training and Validation datasets, adjusted for age, smoking, T-categories, N-categories, and treatment. We used the adjusted hazard ratio (AHR) and recursive-partitioning analysis (RPA) methods to derive cN-classifications that included nodal features that were significant in MVA. We evaluated the performance of stage schemas using the refined cN-classifications (including cN0) against TNM8 for hazard consistency, hazard discrimination, outcome prediction, and sample size balance.
A total of 1898 cN+ (1139 Training; 759 Validation) and 155 cN0 patients were included. iENE+ patients accounted for 37% (710/1898) of cN+ patients. The Training dataset contained 435 (38%) iENE+ and 704 (62%) iENE– cases. iENE positivity increased with higher cN categories: 33% in N1, 43% in N2, and 82% in N3 diseases. Median follow-up was 5.1 years. The iENE+ cohort (vs iENE–) had inferior 5-year OS (69% vs 86%) vs iENE–. MVA confirmed that iENE was prognostic for both Training (Hazard ratio [HR] 2.72) and Validation (HR 1.90) datasets (p<0.001). RPLN and LN number (5+ vs 1-4) were also significant in univariable analysis but non-significant in MVA. We proposed to reclassify iENE+ cases one-stratum higher than that of the existing TNM8 cN, while iENE– either remained as TNM8 cN categories based on AHR model. To evaluate the performance of the stage schemas on OS and DFS, we reassembled cN+ and cN0 patients to create various stage groupings combining the T-categories with TNM8-N, AHR_N, and RPA_N schemas, respectively. The performance of stage grouping with AHR_N and RPA_N were similarly better than that of TNM8 for OS in both Training (1.23 and 1.17 vs 3.03) and Validation (0.85 and 0.37 vs 4.06) datasets. The 5-year OS for the entire cohort by TNM8 stage I/II/III were 89%/80%/62%, by AHR stage I/II/III were 92% vs 83% vs 64%, and by RPA stage I/II/III were 91%/81%/63%. These survival changes are evident in the relocation of higher risk disease with iENE to the higher stage groupings defined by the AHR_N and RPA_N models.
This ICON-N-HPV+ study confirms the prognostic importance of iENE. We propose that iENE+ patients be reclassified one-stratum higher in the N categorization (similar to pN classification in HPV-negative head and neck cancer) while iENE– cases could either remain as per TNM-8 cN (AHR cN schema) or be amalgamated to form a new N1 (RPA cN schema) with consequent modification of the TNM stage groups. Since the AHR2 cN schema retains the TNM8 cN framework and is also consistent with current pN classification for non-viral head and neck cancer, it may be a more practical and less complicated proposal. T1-2_cN+_iENE– patients are potentially more appropriate candidates for deintensification trials while new strategies are needed to address the high risk of disease progression in iENE+ patients.