This study shows a major decrease in radiation-induced dermatitis during treatment with HF compared to CF radiation therapy, with a threefold reduction in the risk of dermatitis. The analyse of effect of HF vs CF treatment within each strata of the baseline variables, revealed a decrease in dermatitis in HF compared to CF group, regardless of treatment site (breast or chest wall), treatment technique (IMRT or conformal), energy type, smoking status, when the treatment included a boost, in the absence of axillary irradiation, and with use of electrons. Conversely, there was no significant difference between HF and CF radiation therapy, when treatment did not include a boost, did not use electrons, or in case of axillary irradiation. The investigation of predictive factors of toxicity in the HF group was exploratory, given the small number of patients, but nonetheless revealed a significant relationship between the irradiation technique and irradiation of the chest wall or breast. The percentage of HF radiotherapy in patients with RNI in our cohort increased between early 2016 and the January 2022. The increase was not linear, and an abrupt change was visible that coincided with the start of the COVID-19 pandemic, rising from around 10% to more than 30% in a few months.
Most studies reporting acute toxicity involved patients who did not have RNI. In spite of this difference, the incidence of toxicity reported in these studies was broadly comparable to that observed in the present HYPOBREAST study. Indeed, the incidence of grade 2 or higher acute dermatitis in patients treated with a HF regimen varied from 5 to 27% in most studies, which is consistent with the 10.7% incidence found in our study. In all these studies, the comparison between the HF and CF groups showed a 2 to 3-fold decrease in acute skin toxicity with hypofractionation, which is also comparable to our results [8, 9, 15,16,17]. Arsenault et al. went further by also analyzing the duration of acute skin injury, and showed that both the peak and the duration of toxicity were reduced by HF compared to CF [8]. The Beijing trial recruited exclusively patients treated on the chest wall with RNI. This trial analyzed skin toxicities by grouping grade 1 and 2, which makes it difficult to compare with our data; however, they found no significant difference between the CF and HF groups [18]. A recent review and meta-analysis of studies that included post-mastectomy irradiation also failed to show any significant difference between CF and HF groups [19]. The START B trial enrolled 7.4% of patients with RNI, and reported no major skin toxicity in this population [4].
Regarding the predictive factors of toxicity in patients treated with HF therapy, no study to date has investigated the predictors of acute dermatitis in this specific population. However, some studies have assessed the risk factors for toxicity in a NF population, and have reported that IMRT is a technique that can reduce skin toxicity, which is consistent with our findings [20, 21]. On the other hand, we describe a significant relationship between acute dermatitis and irradiation of the chest wall or breast, with lower toxicity when the chest wall is irradiated. It has been reported that severe acute cutaneous toxicities are more often observed in the axillary and infra-mammary folds, via a self-bolus effect [22]. Self-bolus effect is the removal of skin sparing effects of megavoltage radiation beams due to build-up of skin folds. It therefore seems logical that fewer grade 2 or higher acute skin toxicity events would be observed in patients receiving post-mastectomy irradiation.
In our study, the use of hypofractionation was different between centers. We therefore included the center in the propensity score, to limit the potential confounding bias that this difference might have generated in our analysis. This difference highlights the heterogeneity in the use of hypofractionated radiotherapy, which has been previously described in the literature. Indeed, Prades et al. attempted to understand the variation in hypofractionation use and the clinical and organizational factors influencing the OR decision [23]. They described clinical factors such as age, indication for chemotherapy, left side, indication for NIR, large breast, chest irradiation, indication for boost, or certain histological subtypes. The authors conclude that the clinical factors cited have little basis in scientific evidence and that factors related to the management of radiotherapy services play a major role. A more recent study, by Ratosa et al., sought to describe the fractionation preferences of radiation oncologists across Europe. Only 28.7% preferred a hypofractionated regimen when irradiating the lymph nodes, while 29.6% preferred a hypofractionated regimen when irradiating the chest wall after mastectomy [24]. The authors also described the reasons influencing the decision to use hypofractionation. The most frequently cited reasons were young age, lymph node irradiation, post-mastectomy indications and breast reconstruction, especially as these are subgroups of patients less represented in the literature. To a lesser extent, organizational aspects and financial issues also had an influence. These study, as well as our cohort, confirms the difficulties of implementing hypofractionated regimens in clinical routine. Indeed, although hypofractionation has many advantages such as patient convenience, accessibility, reduction of waiting lists, better use of limited resources, cost effectiveness, it is still not widely used despite a high-level concerning effectiveness and safety of this approach. As pointed out by Ratosa et al., one of the disadvantages of hypofractionation in some countries is the financial and reimbursement issue. An ESTRO-HERO analysis of reimbursement in Europe, by Lievens et al., highlighted the variability of reimbursement for radiotherapy treatments and the existence of systems that are not adapted to the recent evolution of radiotherapy, such as hypofractionation, and therefore the need to discuss new reimbursement strategies that would allow radiotherapy department to follow evidence-based treatment without being financially disadvantaged [25].
The absence of data concerning breast volume could be a source of potential bias. Indeed, breast size is a risk factor for dermatitis, as previously described in patients treated by conventional fractionated radiotherapy [26, 27]. We cannot exclude the possibility that the breast volume modifies the attitude of the radiation oncologist, and therefore, that patients with a larger breast volume may be more represented in the normofractionated population, which could artificially increase the difference between the 2 groups. We conducted a sensitivity analysis on a sub-cohort with more data (BMI and smoking status). Patients in the sub cohort were less likely to receive axillary irradiation, more likely to receive a boost and to be treated electrons. These differences may affect the generalizability of the results of the sensitivity analysis to the total cohort. However, axillary irradiation did not seem to be related with acute toxicity in our study. Electrons use and boost were more frequent in the sub cohort and could therefore cause greater toxicity in the sub cohort. Despite this, the toxicity reduction with HF, compared to CF radiotherapy, was even more important in the sub cohort. Finally, this sensitivity analysis showed that taking smoking status or BMI into account did not alter the results, and therefore the strength of association is such that even if possible confounding bias existed that was not taken into account, it would not change the final result. In addition, we know that the use of high-energy photons occurs when breast volume is high, and we have shown that hypofractionation was less toxic even when high-energy photons were used. In their study in a population receiving hypofractionated radiotherapy, Janssen et al. did not find any link between breast- or boost-volume, and acute and late toxicity [15]. Corbin et al. compared hypofractionated with conventional fractionated radiotherapy in a large-breasted population [28], and concluded that, in obese and large-breasted populations, there was no increase in acute skin toxicity with the use of hypofractionation. These studies therefore reinforce our conclusion that hypofractionation reduces acute cutaneous toxicities in any population.
Despite clinical trials proving the efficacy and safety of hypofractionated schedules, less than half of patients are treated with hypofractionated schedules. It seems necessary to promote this form of treatment, both for the comfort of patients and to enhance the accessibility of radiotherapy treatments by reducing costs [29, 30]. Moreover, our study demonstrates the feasibility of using real-life data from systematic computerized data collection to conduct large-scale Phase IV studies. Indeed, real-life studies are essential to provide effectiveness data, which complement the efficacy data generated by randomized studies. These two types of study are complementary, because they provide different types of information to clinicians. The development of systematic data collection should be encouraged to collect more population-based data, and should also be extended to include dosimetric data, with a view to strengthening the methodology of real-life studies, and better informing clinicians for their daily practice.