Purpose: To explore valuable predictors for mediastinal lymph node metastasis in non-small cell lung cancer (NSCLC) patients, we analyzed the potential roles of standardized uptake value (SUV)-derived parameters from preoperative 18F-FDG PET/CT combined with clinical characteristics.
Methods: Data from 224 NSCLC patients who underwent preoperative 18F-FDG PET/CT scans in our hospital were collected. Then, a series of clinical parameters including SUV-derived features [SUVmax of mediastinal lymph node and primary-tumor SUVmax, SUVpeak, SUVmean, metabolic tumor volume (MTV) and total lesion glycolysis (TLG)] were evaluated. The best possible cutoff points for all measuring parameters were calculated using receiver operating characteristic curve (ROC) analysis. Predictive analyses were performed using a logistic regression model to determine the predictive factors for mediastinal lymph node metastasis in NSCLC and lung adenocarcinoma patients. After multivariate model construction, data from another 100 NSCLC patients were recorded. Then, 224 patients and 100 patients were enrolled to validate the predictive model by the area under the receiver operating characteristic curve (AUC).
Results: The mediastinal lymph node metastasis rates in 224 patients for model construction and 100 patients for model validation were 24.1% (54/224) and 25% (25/100), respectively. It was found that SUVmax of mediastinal lymph node, primary-tumor SUVpeak ≥ 2.49, primary-tumor SUVmax ≥ 4.11, primary-tumor SUVpeak ≥ 2.92, primary-tumor SUVmean ≥ 2.39, primary-tumor MTV ≥ 30.88 cm3, and primary-tumor TLG ≥ 85.53 were more prone to mediastinal lymph node metastasis through univariate logistic regression analyses. The multivariate logistic regression analyses showed that the SUVmax of mediastinal lymph nodes (≥ 2.49: HR = 7.215, 95% CI: 3.326 - 15.649), primary tumor SUVpeak (≥ 4.11: HR = 5.717, 95% CI: 2.094 - 15.605), CEA (≥ 3.94 ng/ml: HR = 2.467, 95% CI: 1.182 - 5.149), and SCC (< 1.15 ng/ml: HR = 4.795, 95% CI: 2.019 - 11.388) were independent predictive factors for lymph node metastasis in the mediastinum. SUVmax of the mediastinal lymph node (≥ 2.49: HR = 8.067, 95% CI: 3.193 - 20.383), primary-tumor SUVpeak (≥ 4.11: HR = 9.219, 95% CI: 3.096 - 27.452), and CA19-9 (≥ 16.6 U/ml: HR = 3.750, 95% CI: 1.485 - 9.470) were significant predictive factors for mediastinal lymph node metastasis in lung adenocarcinoma patients. The AUCs for the predictive value of the NSCLC multivariate model through internal and external validation were 0.833 (95% CI: 0.769- 0.896) and 0.811 (95% CI: 0.712 - 0.911), respectively.
Conclusion: High SUV-derived parameters (SUVmax of mediastinal lymph node and primary-tumor SUVmax, SUVpeak, SUVmean, MTV and TLG) might provide varying degrees of predictive value for mediastinal lymph node metastasis in NSCLC patients. In particular, the SUVmax of mediastinal lymph nodes and primary tumor SUVpeak could be independently and significantly associated with mediastinal lymph node metastasis in NSCLC and lung adenocarcinoma patients. Internal and external validation confirmed that the pretherapeutic SUVmax of the mediastinal lymph node and primary-tumor SUVpeak combined with serum CEA and SCC can effectively predict mediastinal lymph node metastasis.