A Predictive Model for Lymph Node Involvement with Malignancy on PET/CT in Non-Small-Cell Lung Cancer
Malcolm D Mattes1, Wolfgang A Weber, Amanda Foster
1*Department of Radiation Oncology, West Virginia University, Morgantown, West Virginia; Departments of †Radiology and ‡Radiation Oncology, Memorial Sloan-Kettering Cancer Center, New York, New York; §Department of Radiation Oncology, New York Methodist Hospital, Brooklyn, New York; and Departments of ‖Epidemiology and Biostatistics and ¶Surgery, Memorial Sloan-Kettering Cancer Center, New York, New York.
Accurate assessment of lymph node (LN) involvement in non-small-cell lung cancer is crucial for staging. Adenocarcinoma histology, higher LN SUVmax, and higher LN risk category independently predict malignant LN involvement.
Area of Science:
- Oncology
- Radiology
- Pathology
Background:
- Accurate assessment of lymph node (LN) involvement is critical for staging and managing non-small-cell lung cancer (NSCLC).
- Positron emission tomography/computed tomography (PET/CT) is a key imaging modality for detecting metastatic disease in NSCLC patients.
Purpose of the Study:
- To identify independent tumor and imaging characteristics associated with malignant LN involvement in NSCLC.
- To develop a predictive model for LN involvement risk in NSCLC.
Main Methods:
- Retrospective analysis of 172 NSCLC patients who underwent PET/CT before LN biopsy.
- Logistic regression analysis of 504 anatomically defined, pathology-confirmed LNs to assess associations between nodal involvement and clinical/imaging variables.
- Development of a nomogram based on significant predictors for clinical application.
Main Results:
- Univariate analysis showed adenocarcinoma histology, high LN risk category, larger LN short-axis dimension, and higher LN SUVmax correlated with nodal involvement.
- Multivariate analysis revealed adenocarcinoma histology, high LN risk category, and higher LN SUVmax as independent predictors of nodal involvement.
- A nomogram demonstrated excellent concordance (0.95) between predicted and observed results.
Conclusions:
- Adenocarcinoma histology, higher LN SUVmax, and higher LN risk category are independent predictors of malignant LN involvement in NSCLC.
- These factors can be utilized in a predictive model to accurately assess the risk of nodal metastasis.
- The findings support the use of PET/CT imaging characteristics in conjunction with histology for improved NSCLC staging.
