Prospective intra/inter-observer evaluation of pre-brachytherapy cervical cancer tumor width measured in TRUS and MR imaging

Abstract

Background: Ultrasound (US) imaging has been proved as an excellent diagnostic tool in gynecology and, due to its wide availability and limited cost, is under intense investigation as base for dose adaptation in cervical cancer brachytherapy. Purpose of this work is to test inter/intra-observer uncertainties between magnetic resonance (MR) and trans-rectal ultrasound (TRUS) imaging in defining maximum tumor width before first brachytherapy (BT) application in a prospective cohort of cervical cancer patients undergoing image-guided adaptive brachytherapy (IGABT). Methods: One hundred ten consecutive cervical cancer patients treated between 2013 and 2016 were included. Before the first BT implant patients underwent MR and TRUS scan with no applicator in place. Images were independently analyzed by three examiners, blinded to the other's results. With clinical information at hand, maximum tumor width was measured on preBT TRUS and MR. Quantitative agreement analysis was undertaken. Intra-class correlation coefficient (ICC), Passing-Bablok and Bland Altman plots were used to evaluate the intra/inter-observers measurement agreement. Results: Average difference between tumor width measured on MR (HRCTVMR) and TRUS (HRCTVTRUS) was 1.3 ± 3.2 mm (p < 0.001); 1.1 ± 4.6 mm (p = 0.01) and 0.7 ± 3 mm (p = 0.01). The error was less than 3 mm in 79, 82 and 80% of the measurements for the three observers, respectively. Intra-observer ICC was 0.96 (CI95% 0.94-0.97), 0.93 (CI95% 0.9-0.95) and 0.96 (CI95% 0.95-0.98) respectively. Inter-observer ICC for HRCTVMR width measures was 0.92 (CI95% 0.89-0.94) with no difference among FIGO stages. Inter-observer ICC for HRCTVTRUS was 0.86 (CI95% 0.81-0.9). For FIGO stage I and II tumors, ICC HRCTVTRUS values were comparable to respective HRCTVMR ICC values. For larger tumors HRCTVTRUS inter-observer ICC values were lower than respective HRCTVMR although remaining acceptable. Conclusions: Our results suggest that TRUS is equivalent to MR in assessing preBT tumor maximum width in cervical cancer FIGO stage I/II. In more advanced stages TRUS seems to be slightly inferior to MR although maintaining a good agreement to gold standard imaging.