Clinical validation of a workflow interfacing radiation oncology lung stereotactic body radiotherapy treatment planning and diagnostic radiology

Committee/Faculty

Presenting Author(s)

RH

Ryan J. Hutten, n/a

MD
Department of Radiation Oncology, Huntsman Cancer Hospital, University of Utah
Salt Lake City, Utah, United States

Disclosure: I do not have any relevant financial / non-financial relationships with any proprietary interests.

Co-Author(s)

GN

Geoff Nelson

PhD
Department of Radiation Oncology, Huntsman Cancer Hospital, University of Utah
Salt Lake City, Utah, United States

Disclosure: Disclosure information not submitted.
VS

Vikren Sarkar

PhD
Department of Radiation Oncology, Huntsman Cancer Hospital, University of Utah
Salt Lake City, Utah, United States

Disclosure: Disclosure information not submitted.
SJ

Skyler B. Johnson

MD
Department of Radiation Oncology, Huntsman Cancer Hospital, University of Utah
Salt Lake City, Utah, United States

Disclosure: Disclosure information not submitted.
RT

Randa Tao

MD
Department of Radiation Oncology, Huntsman Cancer Hospital, University of Utah
Salt Lake City, Utah, United States

Disclosure: Disclosure information not submitted.
YH

Ying Hitchcock

MD
Department of Radiation Oncology, Huntsman Cancer Hospital, University of Utah
Salt Lake City, Utah, United States

Disclosure: Disclosure information not submitted.
JC

Jessica Chan

MD
2. Department of Radiology and Imaging Sciences, University of Utah
Salt Lake City, Utah, United States

Disclosure: Disclosure information not submitted.
JS

Joyce Schroeder

MD
2. Department of Radiology and Imaging Sciences, University of Utah
Salt Lake City, Utah, United States

Disclosure: Disclosure information not submitted.
KK

Kristine Kokeny

MD
Department of Radiation Oncology, Huntsman Cancer Hospital, University of Utah
Salt Lake City, Utah, United States

Disclosure: Disclosure information not submitted.

Purpose: Following lung stereotactic body radiotherapy (SBRT), multiple complex changes can occur that make radiologic differentiation of local recurrence (LR) from radiation induced lung injury (RILI) challenging. Technological barriers within healthcare systems can further complicate the interpretation of follow-up diagnostic imaging studies. A major barrier is that radiation treatment planning is conducted using software systems not accessible to diagnostic radiologists, preventing direct comparisons between radiation planning CT scan with dose information and subsequent diagnostic images. Herein, we propose an automated workflow for interfacing radiation dose information from lung SBRT treatments with picture archiving and communication system (PACS) accessible by diagnostic radiology.

Methodology: In this IRB approved, single-center, retrospective study, we identified patients with primary early stage (T1-T2,N0) non-small cell lung cancer (NSCLC) treated with SBRT to a dose of 48-55 Gy in 3-5 fractions between 2017 and 2020. All patients had available pretreatment and 6-month post-treatment diagnostic CT chest. An automated workflow was developed in MIM Maestro (MIM Software Inc. Cleveland OH) to produce biologically effective dose (BED) maps (alpha/beta = 10) from SBRT treatment plans. Simplified dose maps displaying high-dose (BED10 = 100 Gy) and low-dose (BED10 = 24.48 Gy) isodose lines were exported to an anonymized study folder in PACS. Two thoracic diagnostic radiologists reviewed pre- and post-treatment images with and without BED-dose maps available, completing a standardized questionnaire. Presence of RILI and LR were reported as binary variables (yes/no). Confidence in each was evaluated on a continuous scale from 1 to 10, with 10 indicating certainty. Recommendation for additional diagnostic workup was collected as a binary variable (yes/no). The clinical impact of the BED-dose map was evaluated by comparing the two evaluation groups using paired 2-sample t-tests.

Results: Among 30 patients, 86% were stage T1a-T1c, with 53% adenocarcinoma and 40% squamous cell carcinoma. Evaluations of each cohort were completed by each reviewer resulting in 120 total data points for analysis. The most common SBRT dose was 50 Gy in 5 fractions (BED10=100 Gy). The addition of the BED-dose map did not significantly change identification rates of RILI (92% vs. 95%, p=0.32) but did significantly decrease the identification rates of LR (22% vs. 8%, p=0.003). The addition of BED-dose maps significantly increased confidence in calling RILI (7.75 vs. 8.82, p=0.004) and LR (5.5 vs 6.6, p=0.005). Recommendation for additional workup was not significantly different (10% vs 7%, p=0.41).

Conclusions: We demonstrated the feasibility and clinical utility of a workflow generating simplified radiation dose maps that are viewable within PACS for diagnostic radiology review. The availability of BED-dose maps statistically decreased the proportion of radiology reads concerning for LR and statistically increased radiologist confidence in identifying both RILI and LR.

References: