Objectives To determine the correlation between surgeon radiology assessment with laparoscopic scoring in patients with newly diagnosed advanced ovarian cancer.
Methods Following IRB approval, 14 gynecologic oncology surgeons from a single institution performed a blinded review of radiology imaging from 20 patients with advanced ovarian cancer. All patients previously underwent laparoscopic scoring using a validated scoring method from April 2013 to December 2017. Surgeons viewed contrasted CT imaging reports and images in a blinded fashion and recorded PIV scores using the validated scoring method. Linear mixed models (LMM) were conducted to calculate the correlation between radiology and laparoscopic score for each surgeon and the group, and the inter-class correlation (ICC) was calculated.
Results The kappa inter-rater agreement was −0.017 (95% CI 0.023 to −0.005), indicating low inter-rater agreement between radiology and actual laparoscopic score. The ICC was 0.06 (0.02–0.21), indicating that surgeons do not score the same across all images. When using a PIV cutoff of 8, the probability of agreement between radiology and actual laparoscopic score was 0.56 (95% CI 0.49–0.73). When looking at disease site subscales, the probability of agreement was (95% CI): peritoneum 0.57 (0.51–0.62), diaphragm 0.54 (0.48–0.60), mesentery 0.51 (0.45–0.57), omentum 0.61 (0.55–0.67), bowel 0.54 (0.44–0.64), stomach 0.71 (CI 0.65–0.76), and liver 0.36 (CI 0.31–0.42).
Conclusions Surgeon radiology review did not highly correlate with actual laparoscopy findings. By subscale, the best agreement is seen when evaluating for stomach involvement, and the worst with liver involvement. Our study highlights the benefits of laparoscopic assessment to determine resectability over radiology alone.
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.