Towards an integrated approach for tailored radiotherapy treatment in breast cancer

One Belgian women in nine will develop breast cancer before the age of 75. The treatment regimen consists of a combination of a locoregional approach with surgery and/or radiotherapy (RT) and a systemic approach with chemotherapy, hormonal therapy and targeted therapy. Adjuvant RT to the breast, chest wall or regional lymph nodes has shown to significantly reduce recurrences and improve survival. Improvements in screening, diagnosis and therapy have led to ten year survival rates exceeding 70 percent. The increasing number of breast cancer survivors has imposed a growing demand to identify the long-term toxicity of breast cancer treatment and its impact on quality of life (QOL). A philosophy change has thereby slowly been introduced for treatment elaboration evolving from ‘the more, the better’ and the maximal tolerable treatment, to the minimal effective treatment. Breast cancer treatment is moving from a one-size-fits-all therapy to a patient-tailored approach.

An individualized approach ideally integrates pathological, histological, molecular and genetic tumour aspects, the patient’s characteristics and the patient’s preferences. There are two main pathways for locoregional therapy. Firstly, there is breast-conserving therapy (BCT), including breast-conserving surgery (BCS) followed by RT; and secondly there is mastectomy (ME) with or without adjuvant RT; both with axillary nodal assessment and treatment. Level-I evidence has shown equivalence of these two strategies in terms of survival, making BCT the preferred treatment approach. In this thesis project, several directions for a patient-tailored approach were therefore explored.

Within PART I of this project we aimed to tailor the RT indications. Two directions can be followed: either towards treatment intensification or towards treatment de-escalation.

On the one hand, even though survival rates are high, mortality rates remain up to 30 %. Identifying patients with a poor prognosis can thereby form the basis for treatment intensification. In Chapter 3, we aimed to define a patient group in whom RT treatment intensification would be beneficial. Triple-negative breast cancer (TNBC), comprises around 15 % of breast cancers and has worse outcomes compared to other breast cancer subtypes. Because of the poorer outcome, an intensified locoregional approach could be suggested. Therefore two strategies can be proposed: either adjuvant RT or more radical surgery. In Chapter 3 we have retrospectively compared 439 patients treated with BCT, ME only and ME+RT. Locoregional recurrences were significantly higher in the BCT and the ME only group compared to the ME+RT group. Breast-cancer specific survival in our cohort was significantly lower in the ME only group compared to the BCT group and the ME+RT group. Our data thus support the strategy of incorporating RT in the locoregional treatment of TNBC patients. Current international guidelines advise post-mastectomy RT (PMRT) in large, node-positive tumours or positive surgical margins regardless of the tumour subtype. Subgroup analysis of our study including only T1-2N0 TNBC patients with negative surgical margins, in whom PMRT would thus not be advised according to current guidelines, did not show any difference between the three groups for locoregional recurrences and statistically just reached significance in favour of adding RT for breast cancer-specific survival. In clinical practice, the current PMRT guidelines are therefore still valid and patients with TNBC should not be regarded as a separate subgroup.

On the other hand, adding RT to BCS decreases recurrence rates by at least 50 % and improves survival. However in very low-risk patients, the gain of adding RT in absolute numbers might be low and RT treatment can be de-escalated or abolished. In chapter 4 to 6, we wanted to help select patients in whom de-escalation of treatment is valid. Several predictive algorithms have been developed to assist care providers and patients with the therapy decision making in breast cancer treatment. Before its widespread implementation as a useful clinical tool, an objective and thorough validation of any predictive algorithm is of critical importance.

The purpose of the study in Chapter 4 was to externally validate the web-based predictive IBTR! 2.0 algorithm that estimates the 10-year risk of ipsilateral breast tumour recurrence (IBTR), with and without whole-breast irradiation (WBI) after BCS. Therefore, we have retrospectively evaluated 1898 breast cancer cases treated with BCS and RT. The performance of the nomogram was acceptable in the present cohort, slightly lower than the previous reports on smaller patient cohorts, with an overestimation in all risk groups. IBTR! 2.0 can thus be used in practice keeping in mind that absolute current IBTR rates will be lower than predicted.

In Chapter 5, we tested the nomogram developed based on the EORTC boost vs no boost study to predict the gain of adding a boost after WBI by retrospectively analysing 1787 BC cases. The model performed suboptimal and appeared to overestimate the risk of IBTR in general. We therefore concluded not to use the nomogram in our current patient population.

To further investigate who could benefit from a boost to the tumour bed, we have conducted a Cochrane systematic review and meta-analysis of all randomized trials in Chapter 6. We included 5 randomized controlled trials (RCTs) analysing 8325 women. Local control appeared better for women receiving a tumour bed boost compared to no tumour bed boost. The local recurrence rate was thereby decreased by 25 per 1000 women (range 17 to 31 per 1000) at 5 years. Overall survival and disease-free survival did not differ with or without a tumour bed boost. Aesthetic outcome (AO) scored by a panel was better in the no boost group. Late toxicity scored by means of percentage of breast retraction assessment and AO scored by a physician did not differ with or without a tumour bed boost. With the current available data, we were unable to define any subgroups to guide an individualized approach. The absolute size of the benefit of the tumour bed boost is linked to the baseline risk of local recurrence and seems independent of the patient’s age.

The purpose of PART II of the thesis, was to tailor the RT treatment to the individual patient aiming for improved AO and patient-reported outcomes (PROs).

In Chapter 7 we investigated whether the applied boost technique influences the magnitude of the recurrence risk reduction. We have therefore retrospectively evaluated 1879 breast cancer cases treated with an electron boost, a brachytherapy (BT) boost or a photon boost. With a median follow-up of 13 years, we found no difference for the risk of local recurrence between the three groups. The 10-year local recurrence rate in the present study was 2.2 % and thus very low.

Since equal control rates were obtained in the electron group and the BT group, a further clinically relevant issue is how side-effects and PROs in both groups compare. Within this PhD project, we have therefore set up a prospective cohort study following up patients treated with BCT for two years (Chapter 9). Besides objectifying side-effects and PROs with regard to the boost technique, the main focus was to gain knowledge on AO in breast cancer patients treated with BCT. Therefore, we have simultaneously conducted a cross-sectional cohort study (Chapter 8). The purpose of the cross-sectional cohort study in Chapter 8 was to develop a prediction model for late unfavourable AO. We have investigated which clinicopathological and dose-volume metrics of the RT treatment were associated with an unfavourable AO and relevant variables were combined into a prediction model. One-hundred twenty-one BC patients treated with BCT between two and six years ago were included in the analysis and AO was measured using the BCCT.core software based on standardized photographs. Forty-four patients (36 %) developed unfavourable AO. In the optimal multivariable logistic regression model, axillary mymph node dissection (ALND) and seroma were withheld as clinically relevant factors and the volume of the breast receiving 55 Gy (V55) as dose-volume metric.

The first aim of the prospective study in Chapter 9 was to estimate the probability of unfavourable AO two years after RT in patients treated with different boost techniques in the BCT setting. One-hundred and seventy-five patients were prospectively followed-up for two years. Ninety-five patients were treated with an externa beam RT (EBRT) boost and 80 with a BT boost. Two years after RT, 28 % of the patients had unfavourable AO measured with BCCT.core and the results did not significantly differ between patients treated with an EBRT or a BT boost. The results were similar for patient-reported AO (31 %). Further assessed PROs were on QOL and body image and we found no evidence for a difference in PROs between the two boost techniques.

The second aim of the prospective study in Chapter 9 was to gain knowledge on PROs related to RT in the BCT setting. During RT, small fatigue and cognitive impairments were observed along with intermediate body image disturbances. Important to remark is that before the start of RT already, unsatisfactory levels were present for several global, functional, symptom and body image scales. Cancer-related QOL recovered well up to one year after treatment and body image up to two years after RT. However, even though improvements were observed from baseline up to two years after RT, mean patients’ scores indicated persistent unsatisfactory global health status, impaired future perspectives and complaints of fatigue and insomnia. A second finding was that AO scored with the BCCT.core and by the patient correlated well with the body image measures meaning both measurements for AO are valid.

The third aim of the prospective study in Chapter 9 was to test the NTCP model obtained from the cross-sectional study described in Chapter 8. Excluding the patients with unfavourable AO at baseline, those lost to follow-up and one patient that underwent a bilateral ME, the model was validated for 166 patients. The model performed good in the present population.

This thesis project has contributed to the knowledge on treatment individualization. Inherent thereby is that pros and cons of therapy and its side-effects are balanced against each other and that the patient’s voice is integrated in the treatment elaboration.