Objective: To study the accuracy of magnetic resonance imaging (MRI) in lymph node detection in patients with vulva carcinoma.
Methods: Sixty patients with diagnosed vulva carcinoma underwent MRI examination for preoperative evaluation of lymph nodes. MRI images were read independently and retrospectively by two radiologists, both unaware of physical examination and surgery findings. The following characteristics of each lymph node with a short-axis diameter of >or=8 mm were recorded: size (axial, sagittal and coronal); aspect (homogeneous, with fatty center or partial fat); margin (smooth, lobulated/speculated or indistinct); shape: (round, ovoid or elongated). Based on these characteristics, each lymph node was classified as malignant or benign and subsequently each groin was classified as malignant or benign. Histopathology obtained at sentinel node procedure or by inguinofemoral lymphadenectomy was used as reference standard. Per groin sensitivity, specificity, positive and negative predictive values were calculated. Kappa statistics on per groin basis were calculated to express interobserver agreement.
Results: One hundred nineteen groins were examined either by sentinel node procedure or surgery, of which 23 groins were malignant. Sensitivity, specificity, positive and negative predictive values were 52%, 85%, 46% and 87% for observer 1 and 52% 89%, 52% and 89% for observer 2. The interobserver agreement was 104/119 (kappa 0.62), representing good agreement.
Conclusion: At this stage there is no role for standard MRI in evaluating lymph node involvement in patients with vulva carcinoma.