Prediction of lymph node metastasis using the combined criteria of helical CT and mRNA expression profiling for non-small cell lung cancer

Abstract

To improve the diagnostic accuracy of nodal metastasis, we suggest new criteria for the prediction of nodal metastasis with combining CT and mRNA expression profiling. Gene signatures related to nodal metastasis were selected from a microarray using extracted mRNA of 112 patients who underwent surgical resection for non-small cell lung cancer. Included patients were randomized into two groups; the training set (n=79) and the test set (n=33). On the basis of the gene signatures, the chest CTs of the training set of patients were re-analyzed and we set up hypothetical criteria for nodal diagnosis. Thirty-one genes were selected from the mRNA expression profiling to separate the LN-metastasis prediction (+) and LN-metastasis prediction (-) groups. On the basis of these signatures, the criteria of lymph node was adjusted (1) in cases of 'LN-metastasis prediction (+)', mediastinal nodes greater than a 5mm in short axis diameter and detectable hilar nodes were considered as metastatic, and (2) in cases of 'LN-metastasis prediction (-), the conventional size criterion was applied for both mediastinal and hilar lymphadenopathies, except for enlarged nodes along with obstructive pneumonia. The sensitivity and accuracy for the nodal diagnosis were improved from 31% to 85% and 58% to 86%, respectively (p<0.05) by using the combined criteria of CT and the microarray results in the test set as compared to those of CT alone. Prediction of lymph node metastasis using combination of gene signatures and chest CT is superior to the CT-only diagnosis.