Development of a national deep learning-based auto-segmentation model for the heart on clinical delineations from the DBCG RT nation cohort.

BACKGROUND: This study aimed at investigating the feasibility of developing a deep learning-based auto-segmentation model for the heart trained on clinical delineations. MATERIAL AND METHODS: This study included two different datasets. The first dataset contained clinical heart delineations from the DBCG RT Nation study (1,561 patients). The second dataset was smaller (114 patients), but with corrected heart delineations. Before training the model on the clinical delineations an outlier-detection was performed, to remove cases with gross deviations from the delineation guideline. No outlier detection was performed for the dataset with corrected heart delineations. Both models were trained with a 3D full resolution nnUNet. The models were evaluated with the dice similarity coefficient (DSC), 95% Hausdorff distance (HD95) and Mean Surface Distance (MSD). The difference between the models were tested with the Mann-Whitney U-test. The balance of dataset quantity versus quality was investigated, by stepwise reducing the cohort size for the model trained on clinical delineations. RESULTS: During the outlier-detection 137 patients were excluded from the clinical cohort due to non-compliance with delineation guidelines. The model trained on the curated clinical cohort performed with a median DSC of 0.96 (IQR 0.94-0.96), median HD95 of 4.00 mm (IQR 3.00 mm-6.00 mm) and a median MSD of 1.49 mm (IQR 1.12 mm-2.02 mm). The model trained on the dedicated and corrected cohort performed with a median DSC of 0.95 (IQR 0.93-0.96), median HD95 of 5.65 mm (IQR 3.37 mm-8.62 mm) and median MSD of 1.63 mm (IQR 1.35 mm-2.11 mm). The difference between the two models were found non-significant for all metrics (p > 0.05). Reduction of cohort size showed no significant difference for all metrics (p > 0.05). However, with the smallest cohort size, a few outlier structures were found. CONCLUSIONS: This study demonstrated a deep learning-based auto-segmentation model trained on curated clinical delineations which performs on par with a model trained on dedicated delineations, making it easier to develop multi-institutional auto-segmentation models.

Prioritizing clinical trial quality assurance for photons and protons: A failure modes and effects analysis (FMEA) comparison.

BACKGROUND AND PURPOSE: The Global Clinical Trials RTQA Harmonization Group (GHG) set out to evaluate and prioritize clinical trial quality assurance. METHODS: The GHG compiled a list of radiotherapy quality assurance (QA) tests performed for proton and photon therapy clinical trials. These tests were compared between modalities to assess whether there was a need for different types of assessments per modality. A failure modes and effects analysis (FMEA) was performed to assess the risk of each QA failure. RESULTS: The risk analysis showed that proton and photon therapy shared four out of five of their highest-risk failures (end-to-end anthropomorphic phantom test, phantom tests using respiratory motion, pre-treatment patient plan review of contouring/outlining, and on-treatment/post-treatment patient plan review of dosimetric coverage). While similar trends were observed, proton therapy had higher risk failures, driven by higher severity scores. A sub-analysis of occurrence × severity scores identified high-risk scores to prioritize for improvements in RTQA detectability. A novel severity scaler was introduced to account for the number of patients affected by each failure. This scaler did not substantially alter the ranking of tests, but it elevated the QA program evaluation to the top 20th percentile. This is the first FMEA performed for clinical trial quality assurance. CONCLUSION: The identification of high-risk errors associated with clinical trials is valuable to prioritize and reduce errors in radiotherapy and improve the quality of trial data and outcomes, and can be applied to optimize clinical radiotherapy QA.

Reduction in myocardial function and oxygen consumption after chemoradiotherapy in patients with esophageal cancer.

BACKGROUND: Chemoradiotherapy (CRT) may induce myocardial dysfunction, congestive heart failure, and impaired physical performance in patients with esophageal cancer (EC). We aimed to investigate left ventricular (LV) function at rest and during stress, using echocardiography (echo) and a cardiopulmonary exercise (CPX) test both before and immediately after completing CRT. MATERIAL AND METHODS: Consecutive EC patients referred for curative treatment were enrolled. Patients attended either definitive CRT or neoadjuvant CRT with subsequent surgery. The evaluation included cardiac biomarkers, electrocardiogram, echo, and CPX test. The primary endpoint was changes in left ventricular (LV) global longitudinal strain (GLS) at rest. Secondary endpoints were LV ejection fraction (LVEF), LV diastolic function, LVEF and GLS at peak exercise, and maximal oxygen consumption (VO2max). The trial was registered with ClinicalTrials.gov (NCT03619317). RESULTS: Among 47 patients enrolled (94% male; median age 67 years, range 50-86 years), cardiac examinations were performed a median of three days [Interquartile range (IQR (1-5))] before CRT and one day [IQR (0-6)] after CRT. At rest, GLS and LVEF decreased, 17.6 vs. 16.4% and 56.4 vs. 55.1%, respectively (p = 0.004; p = 0.030). Furthermore, an absolute decrease of at least 5% in LVEF and 2.5% in GLS was noted in 21% of the patients. Signs of LV diastolic dysfunction increased from 13 to 21% (p = ns). VO2max significantly decreased; 21.2 ml/kg/min vs. 18.8 ml/kg/min (p < 0.001). CONCLUSION: LV function and physical performance decreased in EC patients after CRT, and the LV systolic reserve capacity declined. This study highlighted that EC treatment was associated with early cardiac side effects, which may have clinical and prognostic implications.

The NARLAL2 dose escalation trial: dosimetric implications of inter-fractional changes in organs at risk.

BACKGROUND: Phase II trials suggested that survival rates for locally advanced lung cancer could be increased by radiotherapy dose escalation. However, results of the phase III RTOG 0617 trial illustrated an imminent risk of treatment-related death. This could be thwarted with strict constraints to organs at risk (OARs) and control of the delivered dose. This study investigates the impact of anatomical changes during radiotherapy on escalated dose distributions used in the Danish NARLAL2 dose escalation trial. MATERIAL AND METHODS: The phase III NARLAL2 trial randomizes patients between a standard and an escalated treatment plan. In the escalated arm, mean doses up to 95 Gy/33 fractions (tumour) and 74 Gy/33 fractions (lymph nodes) are delivered to the most 18fluorodeoxyglucose-positron emission tomography (18FDG PET) active regions. The dose distributions are limited by strict constraints to OARs. For a group of 27 patients, a surveillance scan (sCT) was acquired at fraction 11. The original-escalated treatment plans were recalculated on the sCTs and the impact of inter-fractional changes evaluated. RESULTS: A total of 13 patients (48%) had overdosage of least one OAR. Constraints for the oesophagus, trachea and aorta were violated in 26% of the patients. No overdosage was seen for heart or bronchi. For the connective tissue (all tissue in the mediastinum not identified as OAR or tumour) overdosage was seen in 41% of the patients and for the chest wall in 30% of the patients. The main reason for overdosage was tumour shrinkage. CONCLUSIONS: Anatomical changes during radiotherapy caused one or more OAR constraint violations for approximately half of the patient cohort. The main cause was tumour shrinkage. For lung cancer radiotherapy dose escalation trials, we recommend incorporation of adaptive radiotherapy strategies.

Difference in target definition using three different methods to include respiratory motion in radiotherapy of lung cancer.

INTRODUCTION: Minimizing the planning target volume (PTV) while ensuring sufficient target coverage during the entire respiratory cycle is essential for free-breathing radiotherapy of lung cancer. Different methods are used to incorporate the respiratory motion into the PTV. MATERIAL AND METHODS: Fifteen patients were analyzed. Respiration can be included in the target delineation process creating a respiratory GTV, denoted iGTV. Alternatively, the respiratory amplitude (A) can be measured based on the 4D-CT and A can be incorporated in the margin expansion. The GTV expanded by A yielded GTV + resp, which was compared to iGTV in terms of overlap. Three methods for PTV generation were compared. PTVdel (delineated iGTV expanded to CTV plus PTV margin), PTVσ (GTV expanded to CTV and A was included as a random uncertainty in the CTV to PTV margin) and PTV∑ (GTV expanded to CTV, succeeded by CTV linear expansion by A to CTV + resp, which was finally expanded to PTV∑). RESULTS: Deformation of tumor and lymph nodes during respiration resulted in volume changes between the respiratory phases. The overlap between iGTV and GTV + resp showed that on average 7% of iGTV was outside the GTV + resp implying that GTV + resp did not capture the tumor during the full deformable respiration cycle. A comparison of the PTV volumes showed that PTVσ was smallest and PTVΣ largest for all patients. PTVσ was in mean 14% (31 cm3) smaller than PTVdel, while PTVdel was 7% (20 cm3) smaller than PTVΣ. CONCLUSIONS: PTVσ yields the smallest volumes but does not ensure coverage of tumor during the full respiratory motion due to tumor deformation. Incorporating the respiratory motion in the delineation (PTVdel) takes into account the entire respiratory cycle including deformation, but at the cost, however, of larger treatment volumes. PTVΣ should not be used, since it incorporates the disadvantages of both PTVdel and PTVσ.