Recursive Partitioning Analysis of Mediastinal N2 Lymph Node Involvement with Selected Biological Markers in Operable Non-small Cell Lung Cancer: A Correlative Study

BACKGROUND Expressions of various biomarkers in non-small cell lung cancer (NSCLC) have been linked with the prognosis and involvement of mediastinal lymph nodes. METHODS In this study, we utilized recursive partitioning analysis (RPA) by using P53, c-erb-B2, and P-glycoprotein (PGP) expressions evaluated by immunohistochemistry to estimate retrospectively the likelihood of the occult N2 mediastinal lymph node involvement in patients with operable NSCLC. RESULTS In univariate tests, immunohistochemical staining of the primary tumor for these 3 markers in 61 patients undergoing surgery revealed no direct relationship with the N2 involvement. However, RPA demonstrated in patients aged <75 and with >/=4 mediastinal lymph nodes removed that, high PGP expression frequency (>/=20%) predicted an increased likelihood of the N2 involvement (46.7%, R(2) = 0.25). Univariate nominal logistic regression analysis revealed that RPA group affiliation, and the number of mediastinal lymph nodes resected (logarithmic transformation) were associated with the metastasis to N2 lymph nodes (chi(2) = 17.59, p = 0.0005, and chi(2) = 2.40, p = 0.0654, respectively). Multivariate analysis confirmed that only RPA group affiliation predicted the N2 involvement (chi(2) = 14.63, p = 0.0022). CONCLUSION This study shows for the first time that PGP expression of the primary tumor may help to predict the occult N2 mediastinal lymph node involvement in NSCLC. Thus, further research is required to understand whether PGP expression may aid in the decision process for preoperative mediastinoscopy.