Deep learning-based optimization of field geometry for total marrow irradiation delivered with volumetric modulated arc therapy

BACKGROUND

Total marrow (lymphoid) irradiation (TMI/TMLI) is a radiotherapy treatment used to selectively target the bone marrow and lymph nodes in conditioning regimens for allogeneic hematopoietic stem cell transplantation. A complex field geometry is needed to cover the large planning target volume (PTV) of TMI/TMLI with volumetric modulated arc therapy (VMAT). Five isocenters and ten overlapping fields are needed for the upper body, while, for patients with large anatomical conformation, two specific isocenters are placed on the arms. The creation of a field geometry is clinically challenging and is performed by a medical physicist (MP) specialized in TMI/TMLI.

PURPOSE

To develop convolutional neural networks (CNNs) for automatically generating the field geometry of TMI/TMLI.

METHODS

The dataset comprised 117 patients treated with TMI/TMLI between 2011 and 2023 at our Institute. The CNN input image consisted of three channels, obtained by projecting along the sagittal plane: (1) average CT pixel intensity within the PTV; (2) PTV mask; (3) brain, lungs, liver, bowel, and bladder masks. This "averaged" frontal view combined the information analyzed by the MP when setting the field geometry in the treatment planning system (TPS). Two CNNs were trained to predict the isocenters coordinates and jaws apertures for patients with (CNN-1) and without (CNN-2) isocenters on the arms. Local optimization methods were used to refine the models output based on the anatomy of the patient. Model evaluation was performed on a test set of 15 patients in two ways: (1) by computing the root mean squared error (RMSE) between the CNN output and ground truth; (2) with a qualitative assessment of manual and generated field geometries-scale: 1 = not adequate, 4 = adequate-carried out in blind mode by three MPs with different expertise in TMI/TMLI. The Wilcoxon signed-rank test was used to evaluate the independence of the given scores between manual and generated configurations (p < 0.05 significant).

RESULTS

The average and standard deviation values of RMSE for CNN-1 and CNN-2 before/after local optimization were 15 ± 2/13 ± 3 mm and 16 ± 2/18 ± 4 mm, respectively. The CNNs were integrated into a planning automation software for TMI/TMLI such that the MPs could analyze in detail the proposed field geometries directly in the TPS. The selection of the CNN model to create the field geometry was based on the PTV width to approximate the decision process of an experienced MP and provide a single option of field configuration. We found no significant differences between the manual and generated field geometries for any MP, with median values of 4 versus 4 (p = 0.92), 3 versus 3 (p = 0.78), 4 versus 3 (p = 0.48), respectively. Starting from October 2023, the generated field geometry has been introduced in our clinical practice for prospective patients.

CONCLUSIONS

The generated field geometries were clinically acceptable and adequate, even for an MP with high level of expertise in TMI/TMLI. Incorporating the knowledge of the MPs into the development cycle was crucial for optimizing the models, especially in this scenario with limited data.

Oligometastatic Prostate Cancer Treated with Stereotactic Body Radiation Therapy: The Role of Three-Dimensional Tumour Volume in Patient Survival

AIMS

The definition of oligometastatic prostate cancer (OPCa) is currently based solely on the maximum number of detectable metastases, as there are no validated biomarkers available. The aim of this study was to identify novel predictive factors for OPCa patients who underwent metastases-directed therapy.

MATERIALS AND METHODS

This monocentre, retrospective study included consecutive OPCa patients with a maximum of five metastases in up to two organs, detected with choline- or PSMA-positron emission tomography, who were treated with metastases-directed stereotactic body radiation therapy. Endpoints were overall survival and progression-free survival, assessed with Kaplan-Meier analysis. Univariate and multivariable Cox regression was carried out to evaluate the association between clinical factors and survival outcomes.

RESULTS

Between 2009 and 2021, 163 patients and 320 metastases were treated with 226 stereotactic body radiation therapy courses. The median three-dimensional metastatic tumour volume was 4.1 cm3, with a range from 0.01 to 233.4 cm3. In total, 87 (53.4%), 21 (12.9%) and 55 (33.7%) metastases were classified as cN1, cM1a and cM1b, respectively. The median follow-up was 28.5 months. The rates of overall survival at 1, 3 and 5 years were 89.5% (95% confidence interval 83.4-93.4), 74.9% (95% confidence interval 66.1-81.7) and 57.2% (95% confidence interval 45.8-67.1), respectively. Multivariable analysis showed that overall survival reduced with the increase in three-dimensional total tumour volume (hazard ratio 1.93, 95% confidence interval 1.06-3.52; P = 0.030) and confirmed a significant difference between cN1 versus cM1a-b disease (hazard ratio 1.81, 95% confidence interval 1.01-3.25; P = 0.046). The cut-off value of total volume correlated with the highest risk of death was 20 cm3 (hazard ratio 2.37, 95% confidence interval 1.34-4.18; P = 0.003). The median progression-free survival was 17.8 months, with 1-, 3- and 5-year rates of 63.7% (95% confidence interval 55.4-70.9), 31.5% (95% confidence interval 22.8-40.6) and 24.7% (95% confidence interval 16.0-34.3).

CONCLUSIONS

This study identified three-dimensional total tumour volume and the site of oligometastases as significant predictors of survival in OPCa patients treated with metastases-directed therapy. These parameters can potentially be used to personalised treatment and improve patient outcome.

Total tumor volume as a predictor of survival in patients with multiple oligometastases treated with stereotactic ablative radiotherapy (SABR)

BACKGROUND

Delivering stereotactic ablative radiotherapy (SABR) in patients with multiple oligometastases represents a challenge for clinical and technical reasons. We aimed to evaluate the outcome of patients affected by multiple oligometastases treated with SABR and the impact of tumor volume on survival.

MATERIALS AND METHODS

We included all the patients treated with single course SABR for 3 to 5 extracranial oligometastases. All patients were treated with the volumetric modulated arc therapy (VMAT) technique with ablative intent. End-points of the analysis were overall survival (OS), progression free survival (PFS), local control (LC) and toxicity.

RESULTS

136 patients were treated from 2012 to 2020 on 451 oligometastases. Most common primary tumor was colorectal cancer (44.1%) followed by lung cancer (11.8%). A total of 3, 4 and 5 lesions were simultaneously treated in 102 (75.0%), 26 (19.1%), and 8 (5.9%) patients, respectively. Median total tumor volume (TTV) was 19.1 cc (range 0.6-245.1). With a median follow-up of 25.0 months, OS at 1 and 3 years was 88.4% and 50.2%, respectively. Increasing TTV was independent predictive factor of worse OS (HR 2.37, 95% CI 1.18-4.78, p = 0.014) and PFS (HR 1.63, 95% CI 1.05-2.54; p = 0.028). Median OS was 80.6 months if tumor volume was ≤ 10 cc (1 and 3 years OS rate 93.6% and 77.5%, respectively), and 31.1 months if TTV was higher than 10 cc (1 and 3 years OS rate 86.7% and 42.3%, respectively). Rates of LC at 1 and 3 years were 89.3% and 76.5%. In terms of toxicity, no grade 3 or higher toxicity was reported both in the acute and late settings.

CONCLUSION

We demonstrated the impact of tumor volume on survival and disease control of patients affected by multiple oligometastases treated with single course SABR.

Impact of the Extremities Positioning on the Set-Up Reproducibility for the Total Marrow Irradiation Treatment

Total marrow (lymph node) irradiation (TMI/TMLI) delivery requires more time than standard radiotherapy treatments. The patient's extremities, through the joints, can experience large movements. The reproducibility of TMI/TMLI patients' extremities was evaluated to find the best positioning and reduce unwanted movements. Eighty TMI/TMLI patients were selected (2013-2022). During treatment, a cone-beam computed tomography (CBCT) was performed for each isocenter to reposition the patient. CBCT-CT pairs were evaluated considering: (i) online vector shift (OVS) that matched the two series; (ii) residual vector shift (RVS) to reposition the patient's extremities; (iii) qualitative agreement (range 1-5). Patients were subdivided into (i) arms either leaning on the frame or above the body; (ii) with or without a personal cushion for foot positioning. The Mann-Whitney test was considered (p < 0.05 significant). Six-hundred-twenty-nine CBCTs were analyzed. The median OVS was 4.0 mm, with only 1.6% of cases ranked < 3, and 24% of RVS > 10 mm. Arms leaning on the frame had significantly smaller RVS than above the body (median: 8.0 mm/6.0 mm, p < 0.05). Using a personal cushion for the feet significantly improved the RVS than without cushions (median: 8.5 mm/1.8 mm, p < 0.01). The role and experience of the radiotherapy team are fundamental to optimizing the TMI/TMLI patient setup.

Adaptive Volumetric-Modulated Arc Radiation Therapy for Head and Neck Cancer: Evaluation of Benefit on Target Coverage and Sparing of Organs at Risk

Background: Radiotherapy is essential in the management of head–neck cancer. During the course of radiotherapy, patients may develop significant anatomical changes. Re-planning with adaptive radiotherapy may ensure adequate dose coverage and sparing of organs at risk. We investigated the consequences of adaptive radiotherapy on head–neck cancer patients treated with volumetric-modulated arc radiation therapy compared to simulated non-adaptive plans: Materials and methods: We included in this retrospective dosimetric analysis 56 patients treated with adaptive radiotherapy. The primary aim of the study was to analyze the dosimetric differences with and without an adaptive approach for targets and organs at risk, particularly the spinal cord, parotid glands, oral cavity and larynx. The original plan (OPLAN) was compared to the adaptive plan (APLAN) and to a simulated non-adaptive dosimetric plan (DPLAN). Results: The non-adaptive DPLAN, when compared to OPLAN, showed an increased dose to all organs at risk. Spinal cord D2 increased from 27.91 (21.06–31.76) Gy to 31.39 (27.66–38.79) Gy (p = 0.00). V15, V30 and V45 of the DPLAN vs. the OPLAN increased by 20.6% (p = 0.00), 14.78% (p = 0.00) and 15.55% (p = 0.00) for right parotid; and 16.25% (p = 0.00), 18.7% (p = 0.00) and 20.19% (p = 0.00) for left parotid. A difference of 36.95% was observed in the oral cavity V40 (p = 0.00). Dose coverage was significantly reduced for both CTV (97.90% vs. 99.96%; p = 0.00) and PTV (94.70% vs. 98.72%; p = 0.00). The APLAN compared to the OPLAN had similar values for all organs at risk. Conclusions: The adaptive strategy with re-planning is able to avoid an increase in dose to organs at risk and better target coverage in head–neck cancer patients, with potential benefits in terms of side effects and disease control.

A comprehensive and easy-to-use ECG algorithm to predict the coronary occlusion site in ST-segment elevation myocardial infarction

Background

Several electrocardiogram (ECG) criteria have been proposed to predict the location of the culprit occlusion in specific subsets of patients presenting with ST-segment elevation myocardial infarction (STEMI).

Purpose

The aim of this study was to develop, through an independent validation of currently available criteria, a comprehensive and easy-to-use ECG algorithm, and to test its diagnostic performance and reliability in real-world clinical practice.

Methods

We analyzed ECG and angiographic data from 408 consecutive STEMI patients submitted to primary percutaneous coronary intervention, dividing the overall population into derivation (306 patients) and validation (102 patients) cohorts. In the derivation cohort, we tested &gt;60 previously published ECG criteria, using the decision-tree analysis to develop the algorithm that would best predict the infarct-related artery (IRA) and its occlusion level. We further assessed the new algorithm diagnostic performance and reliability in the validation cohort.

Results

In the derivation cohort, the algorithm correctly predicted the IRA in 91% of cases and both the IRA and its occlusion level (proximal vs. mid-distal) in 73% of cases. When applied to the validation cohort, the algorithm resulted in 90% and 69% diagnostic accuracies, respectively. In a real-world comparative test, the algorithm performed significantly better than expert physicians in identifying the site of the culprit occlusion (p=0.026 vs. best cardiologist and p&lt;0.001 vs.best emergency medicine doctor). The algorithm showed almost perfect reliability (Cohen's kappa 0.86) between a cardiologist and an ambulance paramedic.

Conclusions

Derived from an extensive literature review, this comprehensive and easy-to-use ECG algorithm can accurately predict the IRA and its occlusion level in all-comers STEMI patients.

Funding Acknowledgement

Type of funding sources: None.

Locally Advanced Non-Small Cell Lung Cancer: Clinical Outcome, Toxicity and Predictive Factors in Patients Treated with Hypofractionated Sequential or Exclusive Radiotherapy

Background: This study evaluated the outcome, toxicity and predictive factors in patients unfit for concurrent chemo-radiotherapy (CT-RT) treated with hypofractionated sequential CT-RT or exclusive radiotherapy (RT) for locally advanced non-small cell lung cancer (LA-NSCLC). Methods: We included patients affected by LA-NSCLC (stage IIA-IVA) treated with a total dose of 50–60 Gy in 20 fractions. The primary outcomes were local control (LC), distant metastasis-free survival (DMFS), progression-free survival (PFS) and overall survival (OS). Univariate analysis was used to correlate outcomes with prognostic factors. Results: Between 2011 and 2019, 210 patients were treated, 113 (53.8%) with sequential CT-RT and 97 (46.2%) with exclusive RT. After a median follow-up of 15.3 months, 74 patients (35.2%) had a local progression and 133 (63.3%) had a distant progression. The one-, two- and five-year LC were 73.6%, 55.3% and 47.9%, respectively. At the time of analysis, 167 patients (79.5%) died. The one-, two- and five-year OS were 64.7%, 36% and 20%, respectively. PTV volume correlated with PFS (p = 0.001) and LC (p = 0.005). Acute and late toxicity occurred in 82% and 26% of patients. Conclusions: Albeit with the known limitations of a retrospective and heterogeneous study, our work shows that hypofractionated sequential CT-RT or exclusive RT offer a good local control and toxicity profile and a promising survival rate in LA-NSCLC patients unfit for the concurrent CT-RT scheme.