A comparison of the proposed classifications for the revision of N descriptors for non-small-cell lung cancer.

OBJECTIVES

Several new classifications have been proposed for revision of the N descriptors for non-small-cell lung cancer (NSCLC), but external validation is required. The aim of this study was to validate various newly proposed nodal classifications and to compare the discrimination abilities of these classifications.

METHODS

A retrospective analysis was conducted of 1487 patients who underwent complete resection with systematic lymph node dissection for NSCLC between 2000 and 2008. Four nodal classifications based on the following categories were analysed: zone-based classification (single-zone N1, multiple-zone N1, single-zone N2 and multiple-zone N2), number-based classification (the number of metastatic lymph nodes; 1-2, 3-6 and ≥7), rate-based classification (ratio of the number of metastatic lymph nodes to the total number of resected lymph nodes; ≤15, 15-40 and >40%) and the combination of location- and number-based classification (N1: 1-3, N1: ≥4, N2: 1-3 and N2: ≥4). Concordance (C)-index and net reclassification improvement (NRI) index were used to assess the discrimination abilities of the models.

RESULTS

In multivariate analysis, all of the newly proposed classifications were independent predictors (P < 0.001) of overall survival (OS) after adjustment for significant variables (age, tumour histology and pathological T status). The C-indices of the classifications based on the nodal zone, nodal number, rate and location alongside the number of metastatic lymph nodes were 0.6179, 0.6280, 0.6203 and 0.6221, respectively; however, the differences in the C-indices were statistically insignificant. Compared with the zone-based classification, the NRI for OS of classifications based on the nodal number, rate and location with number were 0.1101, 0.0972 and 0.0416, respectively.

CONCLUSIONS

All four proposed classifications based on the nodal zone, nodal number, rate and the combination of location and number are prognostically valid and could serve as future N descriptors after complete resection of NSCLC. The discrimination ability was not significantly different among the four proposed classifications, although the number-based classification tended to have a higher predictive ability compared with the zone-based classification. Future studies with an in-depth discussion are needed to clarify optimal future N descriptors for NSCLC.