Efficiency of Contemporary Off-line Adaptive Radiotherapy for Lung Cancer

Lung cancer ranks among the most commonly occurring cancer types globally and has the highest mortality rate. Non-small-cell lung cancer (NSCLC) is the predominant form, accounting for more than 85% of cases. Because symptoms often manifest late, many patients present with locally advanced disease (LA-NSCLC) at diagnosis. The cornerstone treatment for unresectable LA-NSCLC consists of a combination of systemic therapy and radiotherapy followed by immunotherapy. Radiotherapy is preceded by a labor- and time-intensive treatment preparation process. This process first requires diagnostic imaging, such as PET-CT, followed by treatment-planning imaging using a dedicated breathing-guided four-dimensional (4D) planning CT. Subsequently, the tumor and surrounding organs at risk (OARs) must be contoured and independently reviewed by radiation oncologists according to internationally recommended guidelines. Commercial convolutional neural networks (CNNs) are currently used to support this contouring process. These models perform voxel-wise classification to label anatomical structures based on prior training data. A widely used commercial CE-marked application for OAR contouring is Syngo.via®. After contouring, dosimetric treatment planning is performed. Plan creation begins with an optimization step followed by a dose calculation, aiming to achieve an acceptable balance between target coverage and OAR sparing. A commonly used CE-marked knowledge-based planning tool that streamlines this process is RapidPlan®. This system provides estimates of achievable dose distributions and guides dosimetrists and medical physicists toward high-quality plan generation. All preparatory steps are executed on the planning CT, which represents only a single anatomical snapshot at one moment in time. Once preparation is complete, the radiotherapy course begins. Treatments are delivered daily over a period of up to seven weeks. Throughout this timeframe, both tumor and intrathoracic anatomical changes may occur, causing the initial contours and treatment plan to no longer match the patient's anatomy on the day of treatment. Adaptive radiotherapy (ART), in which the original plan is modified in response to anatomical changes, can therefore be beneficial. Clinical studies have demonstrated that patients with NSCLC may benefit from frequent treatment-plan adaptations in cases of tumor regression or intrathoracic shifts during the treatment course. Improvements in progression-free survival and overall survival have been reported, and routine mid-treatment offline ART is recommended for LA-NSCLC. However, despite the evidence, very limited number of patients will receive adaptive radiotherapy. The extensive labor associated with contouring and treatment planning remains a major barrier to broad implementation of ART for all eligible patients. Several contemporary tools have the potential to accelerate the adaptive process. In the present study, the investigators intend to prospectively perform mid-treatment adaptations in 30 patients with LA-NSCLC. These patients will undergo a repeated planning CT followed by renewed contouring and treatment planning, mirroring the pre-treatment workflow. This approach enables initial evaluation of the time efficiency of modern contouring and planning tools in an adaptive setting. All tools used are CE-marked and currently applied as part of standard clinical practice on the initial planning CT. Quality checks and manual reviews by medical physicists and physicians will be identical to pre-treatment procedures. For enrolled patients, the resulting treatment plan accounts for anatomical changes and may enhance tumor coverage and OAR protection. A retrospective phase will begin after completion of all treatments. This phase will investigate the feasibility of incorporating state-of-the-art position-verification imaging-namely cone-beam CTs (CBCTs)-as a replacement for repeated planning CTs. If feasible, repeated 4DCT acquisition could be avoided, sparing patients additional imaging doses and reducing the burden of extra appointments. It will be essential to determine whether current contouring and planning tools can be applied effectively to CBCT data. The prospective workflow described above will serve as the benchmark for time efficiency and contouring and planning quality. This comparison will help determine whether transitioning from repeated 4DCTs to CBCTs within an ART workflow is viable and advantageous. In the prospective ART workflow, the repeated 4D planning CT will be acquired, followed by automated OAR contouring using Syngo.via® and manual delineation of tumor lesions. Treatment planning will then be performed with RapidPlan®, which enhances optimization and reduces inter-observer variability. If the resulting adapted plan provides a dosimetric advantage, it will be implemented for the remainder of the radiotherapy course. Plan-verification procedures will mirror those used in pre-treatment planning, including approval by the responsible physicians and medical physicists. Time measurements will be recorded at each workflow step to serve as ground-truth data for the retrospective phase. In the retrospective evaluation, the investigators will assess the performance of a CBCT-based ART workflow. Instead of a repeated 4DCT, daily CBCT images-acquired routinely for position verification-will be used. CBCT image quality for radiotherapy planning will be verified and converted into a synthetic CT using dedicated CE-marked software (MIM®). Synthetic CT generation improves the accuracy of dose calculations. Automated OAR contouring using Syngo.via® and MIM® will then be performed, followed by manual tumor delineation. Subsequent treatment planning will again be carried out using RapidPlan®. This process will evaluate the extent to which CBCT or synthetic CT can support adaptive treatment planning. As in the prospective phase, time registration will occur at all workflow stages.