Interpreting Post-SBRT Lung Imaging: A Comparison of Radiology and Radiation Oncology Impressions.

Stereotactic body radiation therapy (SBRT) is an established treatment option for stage I non-small cell lung cancer. SBRT often causes acute inflammatory changes in the lung as well as chronic fibrosis. These radiologic findings evolve over time and complicate the interpretation of surveillance imaging. Evolving inflammatory/fibrotic changes in the lung can be difficult to distinguish from malignant progression and radiologists and radiation oncologists (ROs) may interpret the same imaging study differently. With increasing access to the electronic medical record, patient's discovery of these disagreements can lead to confusion, fear, or mistrust. We assessed the frequency of discordant radiologic interpretations between radiologists and ROs after lung SBRT and investigated whether certain variables are associated with a higher frequency of inconsistencies.As part of an IRB-approved study, data were collected on patients treated with SBRT at our institution for stage I NSCLC between 2007 and 2014. Every RO encounter that included a CT or PET-CT scan was evaluated. Based on language used, interpretations were scored as "concerning for progression" (CFP), "neutral differential" (ND), or "favor stability or post-radiation changes" (FS). ND required that malignancy was included. Results were considered discordant if one provider interpreted the imaging as CFP and the other as either ND or FS. Results were scored as concordant if radiology interpretations and RO interpretations were both scored as CFP or both scored as either ND or FS. A univariate analysis was conducted exploring the relationship between discordance and the type and timing of scans.During the time interval, 97 patients underwent SBRT for stage I NSCLC. Median follow-up was 15 months. Median (Q1, Q3) number of follow-up scans per patient was 3 (2, 5). Of 273 encounters, 18 (6.6%) had discordant interpretations. A ND was listed more frequently by radiology (18% vs 5%). When ND was listed by radiology (n = 49), RO interpretations favored stability 92% of the time. Local failure occurred in 9 cases, of which 2 patients had discordant reads prior to recurrence. Both were read as CFP by radiology (ND and FS by RO). The remaining 7 were concordantly read as CFP. Neither the type of scan (CT vs PET-CT), year of scan, nor the time from treatment were associated with discordant interpretations on univariate analyses.Surveillance imaging after lung SBRT is often interpreted differently by radiologists and ROs. While outright discordance between impressions at our institution was low, radiologists were more likely to list a differential that included disease progression. In these cases, ROs more often favored stability or post-radiation changes. A recognition of these discrepancies is important when considering patient's access to their medical record.

Duke Scholars

Author Donna Niedzwiecki Biostatistics & Bioinformatics, Division of Biostatistics

Author Jared Dean Christensen Radiology, Cardiothoracic Imaging

Author Christopher Ryan Kelsey Radiation Oncology