Optimization of radiotherapy for breast cancer: Back to basics and beyond

Breast cancer (BCa) is the most common malignancy in women in the world and an important cause of cancer-related mortality. Over the last decades, the prognosis of BCa patients has improved a lot thanks to breast cancer screening programs, the development of newer/better treatments and the increased use of systemic therapies. Many BCa patients are now long-term survivors. Accordingly, treatment-related toxicity should be limited to enhance the quality of life of former BCa patients. In the treatment of BCa, radiotherapy plays an important role. Adjuvant radiotherapy to the breast, thoracic wall and/or loco-regional lymph node areas has shown to significantly reduce local failure and improve overall survival. To diminish the treatment-related toxicity caused by radiotherapy, the doses to the organs at risk (OAR) should be kept as low as possible without compromising the target coverage. Therefore, the aim of this thesis was to investigate different ways to lower the doses to the OAR without compromising the target coverage, especially by going back to the basics and by searching for solutions of still unanswered questions, in order to optimize the radiotherapy process. We have dealt with 3 different topics: (1) treatment indication, (2) position alteration and (3) optimization of the target volume delineation. In the first part of this thesis, we retrospectively identified morphological and molecular risk factors for loco-regional recurrences (LRR) in BCa patients who were treated with mastectomy and adjuvant radiotherapy (chapter 3). Our results demonstrated that triple negative BCa patients, especially in the presence of lymphovascular invasion (LVI), are at increased risk to develop a LRR, despite tri-modality treatment (surgery, radiotherapy and systemic treatment). Therefore these BCa patients may benefit from treatment individualization taken into account morphological and molecular risk factors to intensify the treatment. Also Her-2 enriched BCa came out to be an independent risk factor for LRR in our study. However, in the inclusion period most of the Her-2 positive BCa patients did not receive trastuzumab, because not available at that time. Therefore no clear conclusion can be drawn for the Her-2 enriched BCa subtype in this study. In the second part of this thesis, we evaluated the effect of the treatment position on the doses to the OAR (chapter 4). Three positions were compared: prone position, supine position in free breathing (=standard position in BCa RT) and supine position in deep inspiration (=Breathing Adapted RT). We have shown that for left-sided BCa patients, the heart and the left anterior coronary descending artery (LAD) were significantly better spared in supine position in deep inspiration compared with supine position in free breathing (FB) and prone position. For left- and right-sided BCa patients together, our study showed significantly lower doses to the lungs in prone position compared to the supine position in free breathing and deep inspiration. Since differences in doses to the OAR were seen, while target coverage and dose homogeneity were comparable, we proposed a decision-making flow chart for whole breast irradiation (WBI) taking into account the best of each position. For right-sided BCa patients, we advise to perform WBI in prone position. For left-sided fit BCa patients, we advise WBI in supine position in deep inspiration. For left-sided unfit BCa patients, we recommend to perform WBI in supine position in FB. The third part of this thesis dealt about the optimization of the target volume delineation. For the regional nodal areas, we created, within PROCAB (PROject on CAncer of the Breast), new vessel-based delineation guidelines for the regional nodal irradiation (RNI) in early-stage BCa patients (chapter 5). In contrast to existing guidelines, these guidelines are based on the anatomy of the blood vessels, rather than on bony structures, because the lymphatics in the axilla run along the veins. The advantages of using the veins to guide the target volumes for RNI are multiple. First, it is more patient-tailored and precise, because the anatomy of patients can differ a lot. Second, the treated volumes will not be larger than treated with conventional 2D RT. Third, it is applicable for all treatment positions. Fourth, exclusion of the scapulo-humeral joint (SHJ) to prevent shoulder problems is almost never an issue. The largest difference with the existing guidelines is the cranial border of the supraclavicular region or level IV of the axilla. It is not the caudal border of the cricoid bone anymore, but the top of the subclavian artery arch. Another important difference is that the target volume in cranial-caudal direction is mainly concentrated 5 mm around the veins instead of being defined by muscles and bony structures. For the optimization of the boost volume, we retrospectively investigated the performance of a sequential boost (extra dose to the tumor bed after whole breast irradiation) in 1379 BCa patients (chapter 6). The local control was very good with a 10-years local relapse-free rate of 97.9%. The boost was given with different techniques. In our study, the boost modality was selected for each individual patient using a predefined flowchart, based on the depth of the tumorbed. The three used boost techniques (electrons, photons and brachytherapy) were compared. No difference in local control was observed between those techniques, indicating that no boost technique has been shown to be superior to another using the decision flowchart. With the introduction of more conformal techniques and accelerated partial breast irradiation (APBI), an accurate definition of the boost volume becomes essential. Accordingly, minimizing the boost volume is necessary to ameliorate the cosmetic outcome, while adequate target coverage is mandatory for a good local control. A surrogate to measure the delineation accuracy is the interoberver variability (IOV). Therefore, a prospective study was started to illustrate the problem of IOV for the delineation of the boost volume and to test possible solutions to decrease the IOV. In chapter 7, the use of a pre-operative (pre-op) CT in treatment position was investigated. The purpose of this study was to decrease the interobserver variability and the size of the CTVboost by knowing the localization of the tumor on the pre-op CT. However, our study showed no benefit of the use of a pre-op CT-scan to decrease the IOV or to lower the size of the delineated CTVboost. On the contrary, a highly significant increase of the IOV was observed (i.e. a decrease of the Jaccard Index (JI)). Furthermore, the volume of the delineated CTVboost was not reduced. Other possibilities than using a pre-operative CT-scan to reduce the IOV and to better define the CTVboost should be investigated in the era of more sophisticated and conformal treatment techniques. In chapter 8, anisotropic/asymmetric margin expansion from tumorbed to CTVboost has been compared with isotropic/symmetric margin expansion and the impact on the volume and the IOV of the delineated CTVboost. The use of anisotropic margin expansion from tumorbed to CTVboost isotropic significantly reduced the volume of the delineated CTVboost with a factor 1.9 compared to isotropic margin expansion, but it substantially increased the IOV. This increased IOV reflects the complexity of the manual anisotropic margin expansion and results in more uncertainties about target coverage. The improvement in conformity with an isotropic margin expansion implies less uncertainty. Therefore, as long as an anisotropic margin expansion cannot be performed automatically, the better conformity with an isotropic margin expansion counteracted the preferable smaller size of CTVboost with anisotropic margin expansion. With this PhD thesis, we went back to the basics of the radiotherapy process for BCa and tried to fill in some remaining gaps with the ultimate goal of improving the BCa radiotherapy by lowering the doses to the organs at risk without compromising the target coverage.status: publishe