Local control strategies for management of NSCLC with oligoprogressive disease.

Progresses of systemic treatments in advanced non-small cell lung cancer (NSCLC), such as immune checkpoint blockers (ICB) and targeted therapies, led to the increased incidence of oligoprogressive disease (OPD). The OPD is a subtype of oligometastatic disease (OMD) defined as a progression of a limited number of lesions during systemic treatment exposure. The hypothesis was formulated that local radical treatments (LRT) could eradicate progressive lesions resulting from resistant clones, ultimately leading to systemic treatment sensitivity restoration. Recently published international consensuses and guidelines aim to obtain a uniform definition of OMD NSCLC, to standardize the inclusion of these patients in future clinical trials, as well as their management in daily practice. Although there is no specific definition of OPD, LRT strategies in OPD are supported after reporting promising results. Both retrospective and preliminary prospective randomized data of LRT for patients with OPD NSCLC are encouraging. More clinical and translational data are needed for selecting best scenarios where LRT should be delivered. In this review, we analyze the current available literature on LRT for patients with OPD in advanced NSCLC and discuss about future trial design and challenges.

Safety Margins for the Delineation of the Left Anterior Descending Artery in Patients Treated for Breast Cancer.

PURPOSE: Cardiac toxicity after breast cancer (BC) radiation therapy is partly due to the large radiation doses to coronary arteries. The left anterior descending artery (LAD) is particularly exposed. A first step in achieving robust dose constraints to the LAD during treatment planning is homogeneous delineation based on guidelines. LAD delineation can be problematic due to heart movements. The aim of the study was to establish a safety margin for delineation of the LAD in patients with BC. METHODS AND MATERIALS: We studied 45 patients with left-sided BC who had an indication for adjuvant radiation therapy between 2015 and 2018. They all underwent cardiac-gated computed tomography scan, as well as planning computed tomography scans with or without contrast agents, to assess LAD diameter and movements. Cardiac-gated computed tomography scan was performed during monitoring of the cardiac cycle. Acquisition was launched immediately after contrast injection (arterial sequence), with deep-inspiration breath hold and use of a beta-receptor blocking agent. By manually reviewing each scan, the LAD positions and diameter were defined at 20 different phases of the cardiac cycle at 5 different sites: ostium (OS), circumflex bifurcation (bfc), first diagonal bfc, second diagonal (D2) bfc, and apex (right coronary anastomosis). RESULTS: Movement of the LAD is maximal at the ostium and then constant overall even when far from its origin. The diameter decreases with the distance from ostium: 4.9 mm (OS), 3.9 mm (circumflex), 3.5 mm (D2), and 3.1 mm (D2). CONCLUSIONS: We suggest using a safety delineation margin consisting of a cylinder with a diameter of 10 mm surrounding the LAD. These findings must be validated in independent series of patients treated for BC.

Interpretable Machine Learning Model for Locoregional Relapse Prediction in Oropharyngeal Cancers.

BACKGROUND: There is no evidence to support surgery or radiotherapy as the best treatment for resectable oropharyngeal cancers with a negative HPV status. Predictive algorithms may help to decide which strategy to choose, but they will only be accepted by caregivers and European authorities if they are interpretable. As a proof of concept, we developed a predictive and interpretable algorithm to predict locoregional relapse at 18 months for oropharyngeal cancers as a first step towards that goal. METHODS: The model was based on clinical and Pyradiomics features extracted from the dosimetric CT scan. Intraclass correlation was used to filter out features dependant on delineation. Correlated redundant features were also removed. An XGBoost model was cross-validated and optimised on the HN1 cohort (79 patients), and performances were assessed on the ART ORL cohort (45 patients). The Shapley Values were used to provide an overall and local explanation of the model. RESULTS: On the ART ORL cohort, the model trained on HN1 yielded a precision-or predictive positive value-of 0.92, a recall of 0.42, an area under the curve of the receiver operating characteristic of 0.68 and an accuracy of 0.64. The most contributory features were shape Voxel Volume, grey level size zone matrix Small Area Emphasis (glszmSAE), gldm Dependence Non Uniformity Normalized (gldmDNUN), Sex and Age. CONCLUSIONS: We developed an interpretable and generalizable model that could yield a good precision-positive predictive value-for relapse at 18 months on a different test cohort.

Optimization and auto-segmentation of a high risk cardiac zone for heart sparing in breast cancer radiotherapy.

INTRODUCTION: Breast radiotherapy is associated with a risk of ischemic heart disease. Limiting left anterior descending coronary artery (LADCA) exposure might possibly reduce coronary risk. However, its manual delineation is poorly reproducible and its auto-segmentation remains unreliable. This study aims to define and characterize a high-risk cardiac zone (HRCZ) as a LADCA surrogate and to implement its auto-segmentation. MATERIALS AND METHODS: Forty breast cancer patients treated with adjuvant IMRT were included. We delineated the LADCA and eight HRCZ, defined as 1 cm-thick cardiac wall segments centered on the LADCA with symmetrical lateral margins defining the HRCZ width (ranging between 1 and 8 cm). We retrieved mean and maximum doses to the LADCA and to the HRCZ and calculated relative dose variations. We constituted an atlas with the HRCZ contours of 20 patients. Based on this latter, a commercial atlas-based auto-segmentation software delineated HRCZ for the remaining 20 patients and performances were evaluated using distance metrics. RESULTS: Relative maximum dose variations were systematically positive and increased with HRCZ width, rising from 7.2% to 112.8% for right-sided irradiation (with a sharp increase above 4 cm), and from 9.5% to 30.4% for left-sided irradiation. Auto-segmentation performances asymptotically improved with HRCZ width: Dice similarity coefficient values were 0.62 for a 3 cm width and 0.69 for an 8 cm width. CONCLUSION: A 3.5 cm-wide HRCZ is a reliable LADCA surrogate for breast radiotherapy. Applying maximum dose constraints to the HRCZ could limit LADCA exposure. Auto-segmentation algorithms can reliably delineate HRCZ.