Commentary on “Number of positive lymph nodes better predicts survival for oral cavity cancer” by Hojun Lee et al

Abstract

The introduction of the 8th edition TNM by AJCC and UICC in 2018 has stimulated numerous publications and commentaries comparing and contrasting the current version with prior versions and alternate ideas. Staging has evolved to serve different purposes including to allow clinical trial and treatment guideline entry points, comparisons of outcomes across time and geography, individual patient prognostication, and to better understand the global burden of disease. The need to understand multiple aspects of staging is necessary as we strive to find the best possible system. Staging is inherently a conservative endeavor. Frequent changes make comparisons hard to assess. Staging is a global affair. Resource limitations and the ability to be used in every clinical environment are critical. Thus, there must be a compromise between the excesses of extreme detail and deficiencies of bland generalities. One must strike a balance between the need for consistency over time against the accumulation of evidence that allows for staging evolution. To develop a statistical approach to interrogate and compare various staging systems, Groome et al proposed four key statistical factors to assess the fundamentals of a successful staging system. These factors are hazard consistency (each patient within the same stage should have similar outcomes), hazard discrimination (each stage group should be different than the one above or below), predictive power (disease‐specific and overall survival), and numerical balance (relatively similar numbers in each stage). Thresholds to initiate a change in a staging system require sufficient high‐quality data from diverse sources, worldwide acceptance of the proposed changes, respect for resource limitations, and attention to the many broad goals of staging. The manuscript by Lee et al serves as a case in point to emphasize these fundamental principles of staging. The three most substantial 8th edition AJCC/UICC changes were the introduction of a novel staging for HPV‐associated oropharyngeal cancers, the addition of depth of invasion (DOI) to the oral cavity T category and the more universal introduction of extranodal extension (ENE) to N category for all nonvirally mediated head and neck cancers. Lee et al examined their center s 10‐year experience with oral cavity cancer to address the latter two changes. Their data confirms the enhanced ability of the 8th edition TNM over the 7th edition to predict disease‐specific and overall survival in oral cavity cancers, especially through the addition of DOI. This finding has also been reported by authors in other countries as well. From Brazil, Matos et al examined 298 patients and found improved prognostic information for both DOI and ENE suggesting an improvement in hazard consistency and discrimination. In over 39000 patients from the United States’ National Cancer Database, Cramer et al found an improved survival concordance index and better numerical distribution. Critiques of staging often focus on ways to enhance particular aspects of staging. Number, laterality, and size of nodes have been a hallmark of N categories since the early days of staging. The 8th edition s introduction of ENE to both clinical and pathological staging of nonvirally mediated cancers of the head and neck is intended to further refine prognosis. Lymph node ratio and absolute lymph node number are additional methods of parsing the N category as explained by Lee. The nodal number is used for pathologic staging of HPV‐associated oropharyngeal cancer in the 8th edition TNM, so certainly, the question being asked by the authors is germane. The problem is determining the ideal numerical cut off of nodes to provide adequate hazard consistency and discrimination. In particular, numerical balance must be sought as a stage grouping that is very small even when excellent hazard discrimination does not allow a nuanced distinction between a significant number of patients to render it useful. Lee s manuscript highlights the need to emphasize several key points. Clinical ENE is a high bar that requires strong clinical evidence of extension of cancer into surrounding structures causing dysfunction supported by strong radiographic evidence. The second is that the extent of ENE has prognostic significance. This observation is important and reflects the head and neck task force s belief that this may be the case. As such, registrars are collecting data on the extent of ENE to help address just this eventuality. Of interest, both the addition of DOI in T characterization and the use of node number and size primarily reflect the volume of disease in predicting outcome. ENE, on the other hand, may reflect the biological nature of the disease in a similar manner to perineural invasion and may serve as a surrogate for biological determinants of behavior thus complementing volume. Of concern, in the introduction, the authors incorrectly state “pathological ENEs are staged up to N3b, regardless of the size or number of metastatic LNs.” They reiterate this mistake in the discussion when they state, “The AJCC system (8th edition) classified ENE‐positive in the highest nodal category (N3b).” In fact, this is true of clinical ENE but not pathological ENE, which was the focus of their study. A single node less than or equal to 3 centimeters with ENE is increased to N2a, not N3b. This point may have led to incorrect data analysis inappropriately placing all ENE nodes into N3b. The authors’ claim in the discussion that “the prognostic significance of ENE in