Background and objective: Glucose and FDG compete for uptake by cancers. Here, we undertook to improve diagnostic accuracy of FDG-PET for determining mediastinal lymph node (LN) status in NSCLC by considering serum glucose level.
Methods: NSCLC patients (n = 70) who underwent curative lung resection and mediastinal LN dissection within 1 month of FDG-PET were enrolled. MaxSUV was calculated using lean body weight and used to determine a new parameter (maxSUV x serum glucose level; maxSUV-GL). Histopathologic LN results were compared with maxSUV and maxSUV-GL values.
Results: Of 71 LN stations whose FDG uptake could be measured, 21 were malignant and 50 benign. MaxSUV of LN had AUC of 0.729 (95% CI: 0.610-0.827) by ROC analysis with sensitivity of 47.6% (10/21), specificity of 94.0% (47/50), and a cutoff value of 3.3. Using maxSUV-GL the corresponding values were; AUC 0.825 (95% CI: 0.716-0.905) and sensitivity 76.2% (16/21), with a cutoff value of 290.4, which represented a significant improvement (P < 0.01) without compromising specificity 88.0% (44/50) (P > 0.05). The exclusion of neo-adjuvant chemotherapeutic and diabetic patients resulted in a similar improvement in diagnostic accuracy.
Conclusion: By considering serum glucose level during FDG-PET using the new parameter maxSUV-GL, sensitivity for malignant mediastinal LN detection is improved.
(c) 2006 Wiley-Liss, Inc.