Diagnostic Performance of MRI for Assessing Parametrial Invasion in Cervical Cancer: A Head-to-Head Comparison between Oblique and True Axial T2-Weighted Images

ABSTRACT

OBJECTIVE: To directly compare the diagnostic performance of true and oblique axial T2-weighted imaging (T2WI) for assessing parametrial invasion (PMI) in cervical cancer. MATERIALS AND METHODS: This retrospective study included 71 women with treatment-naive cervical cancer who underwent MRI that included both oblique and true axial T2WI, followed by radical hysterectomy. Two blinded radiologists (Radiologist 1 and Radiologist 2) independently assessed the presence of PMI on both sequences using a 5-point Likert scale. Receiver operating characteristic (ROC) curve analysis was performed, with a subgroup analysis for tumors sized > 2.5 cm and ≤ 2.5 cm in diameter. Inter-reader agreement was assessed with kappa (k) statistics. RESULTS: At hysterectomy, 15 patients (21.1%) had PMI. For Radiologist 1, the area under the ROC curve (AUC) was greater for oblique axial than for true axial T2WI {0.941 (95% confidence interval [CI] = 0.858–0.983) vs. 0.917 (95% CI = 0.827–0.969), p = 0.027}. The difference was not significant for Radiologist 2 (0.879 [95% CI = 0.779–0.944] vs. 0.827 [95% CI = 0.719–0.906], p = 0.153). For tumors > 2.5 cm, AUC was greater with oblique than with true axial T2WI (0.906 vs. 0.860, p = 0.046 for Radiologist 1 and 0.839 vs. 0.765, p = 0.086 for Radiologist 2). Agreement between the radiologists was almost perfect for oblique axial T2WI (k = 0.810) and was substantial for true axial T2WI (k = 0.704). CONCLUSION: Oblique axial T2WI potentially provides greater diagnostic performance than true axial T2WI for determining PMI, particularly for tumors > 2.5 cm. The inter-reader agreement was greater with oblique axial T2WI.