MSSP447: IS IMAGING RELIABLE?: RADIOLOGY AND PATHOLOGY CORRELATION OF BILIARY DISEASE
Sydney E Dishman, MD; Rathnayaka MKD Gunasingha, MD; Elliot M Jessie, MD; Carolyn E Gosztyla, MD; Matthew J Bradley, MD; Walter Reed National Military Medical Center
Introduction: Cholecystectomies are the most common intraabdominal surgeries performed in the US, but the indications for gallbladder removal are broad, making imaging modalities essential to distinguish underlying disease processes. Given that providers depend on radiologic interpretations to guide management of biliary disease, it is important that findings correlate to ultimate pathology. The aim of this study is to review the accuracy of preoperative imaging as it relates to biliary disease and correlate findings to final pathology reports in an effort to identify areas of improvement.
Methods: All adult patients who underwent a cholecystectomy between 2015 and 2016 at a single institution were identified. Medical records were reviewed for patient demographics, abdominal imaging, surgical diagnosis, and final pathology. Patients were excluded if they had missing data. Simple correlation was determined between variables and qualitative data was analyzed using Chi square test. All analysis was completed in R Version 4.0.3.
Results: Between 2015 and 2016, 210 patients underwent a cholecystectomy at our institution. Of those, 166 patients had abdominal imaging within a year of surgery. Overall, preoperative radiologist impressions were found to correlate with pathology only 27.7% of the time. Pathology was significantly discordant with radiographic interpretations, with 21.5% of chronic and 50% of acute cholecystitis correctly correlating (X2(3, N=166)=19.5, p<0.0002). This finding was more pronouncedly seen when comparing impressions obtained from right upper quadrant ultrasounds (RUQUS), which was performed in 85% of patients. Pathology significantly affected RUQUS-pathology correlation with only 13.7% of chronic and 28.5% of acute cholecystitis correctly correlating (X2(3, N=141)=35, p<0.0001). Imaging impressions were found to be particularly discordant with pathologic findings if ultrasound was the imaging modality obtained in closest proximity to operative management (19.4%) or if the imaging was obtained in the subacute period, from 48 hours to 30 days prior to surgery (21.2%).
Conclusion: This study revealed a significant discordance between imaging impressions when compared to final pathology. While a majority of patients obtained an abdominal ultrasound preoperatively, this was found to be the least effective modality in predicting final pathology. Furthermore, radiology findings did not correlate in the setting of chronic cholecystitis, one of the most common indications for cholecystectomy. Interventions specifically directed at standardizing the interpretation of preoperative imaging may help alleviate this discrepancy.