Here we show a low absolute risk for a subsequent iIBC at 10-year after a diagnosis and treatment of primary DCIS without invasive breast cancer. With a median time to iIBC of 4.8 years and median follow-up of 8.2 years from patients diagnosed with DCIS from 2005 to 2015, the cumulative incidences of subsequent iIBC are 3.1% after BCS + RT, 7.3% after BCS alone and 1.6% after MST. Although absolute risks of iIBC are low in patients treated for DCIS by either BCS and or BCS + RT, the risk remained higher for patients treated by BCS alone compared to patients treated with BCS + RT for at least 10 years after DCIS diagnosis. Compared to our previous study of van Seijen et al. [15], which included 10,045 primary DCIS patients diagnosed from 1989 to 2004 the current study reports lower absolute risks for iIBCs for the different treatment strategies for primary DCIS. Van Seijen et al. reported 10-year cumulative incidences of 5.2% after BCS + RT, 13.9% after BCS alone and 1.1% after MST with a median follow-up of 15.7 years after diagnosis. Comparing the 10 years cumulative incidences of our previous study to the current study, a reduced risk of approximately 50% for the different treatment strategies, with exception of the MST treated group, is demonstrated. In addition, trends of decreasing hazard ratios over time in the current study were also seen, similar to those reported by van Seijen et al. [15]. The current study more accurately reflects the daily practice in managing DCIS nowadays, since patients included in this study were diagnosed from 2005 to 2015. Current practice comprises a fully implemented Dutch breast cancer screening program and the addition of RT in standard care for DCIS in case of BCS [21]. Luijten et al. [22], demonstrated the patterns of treatment in DCIS patients over time since the introduction of breast cancer screening in the Dutch population. They showed that use of BCS increased from 47.7% in 1995–1996 to 72.7% in 2017–2018. Also, a sharp rise in the use of adjuvant radiotherapy in patients treated with BCS was observed, from 28.9% in 1995–1996 to almost 90% in 2011–2012, followed by a drop to 74.9% in 2017–2018. The addition of radiotherapy could be an explanation for the lower absolute risks for subsequent iIBC as 86.4% of the patients treated with BCS received adjuvant RT compared to just 49.6% of patients from our previous study [14]. The decline in risk of a subsequent breast event after a diagnosis of DCIS over time has been observed in two earlier studies as well [16, 17]. Halasz et al. reported on 246 consecutive patients who underwent BCS and RT for DCIS from 2001 to 2007 and attributed the risk decline to improved resection margins and better detection in modern era mammography [17]. Subhedar et al. retrospectively reviewed a prospectively collected cohort of 2.996 DCIS patients with a median follow-up of 6.3 years treated with BCS from the years 1978–2010 and observed similar declines in the risk of a subsequent breast event with later year of DCIS diagnosis. They concluded that the decline in subsequent breast events after DCIS could only partially be explained by the increased proportion of screen-detected patients, more clear margins, and the increased use of RT [16]. In our study no information was available regarding resection margins, and since in the Netherlands patients do not receive adjuvant endocrine therapy, this was not a possible factor influencing risk of iIBC. Our study did not take into account the non-invasive recurrences. Although they are clinically of less important, these lesions may have a severe impact on patient. Additionally, we investigated whether cumulative incidences in patients low-grade DCIS versus high-grade DCIS showed strong differences. However, these results showed only marginally and clinically non-significant differences (see Figs. 2 and 3).

This study has several strengths and limitations. A limitation of this study is the potential of confounding by indication, considering that women with less favorable characteristics more probably received more invasive treatment in terms of adjuvant radiotherapy which may have resulted in an underestimation of the difference in iIBC risk between BCS + RT and BCS alone. Furthermore, risk factors for developing iIBC such as primary lesion size and margin status could not be studied since information was not available. However, the magnitude of these risk factors associated with a subsequent iIBC after DCIS is still debated [23, 24].

An important strength of this study is that the included population is reflective of the current management of DCIS since adjuvant RT was incorporated as standard care for DCIS, ensuring a homogeneous study population. Also, over the years, more detailed data have been registered by NCR and PALGA, enabling more complete datasets. For this study both the NCR- and PALGA-data were scrutinized to identify primary DCIS patients, providing a true primary pure DCIS cohort. The nationwide NCR registers all primary DCIS patients in the Netherlands as of 2001 and includes both screen-detected and non-screen detected DCISs. Therefore, this dataset is unique with regard to its size and the robustness of the data due to the comprehensive registration of DCIS and IBC.

In conclusion, we report low absolute risks of iIBC after diagnosis of DCIS. These results are in line with more recently reported declining trends of a subsequent iIBC after DCIS. Possible explanations for this declining trend might be the more frequent use of adjuvant RT, an increased proportion of radical resection, and a higher proportion of screen-detected DCIS. The very low risk of an invasive recurrence observed in this study supports current efforts in active surveillance trials to determine whether it is safe to omit loco-regional treatment in patients with lower-grade (grade I/II) DCIS [25,26,27]. For high-grade DCIS the results of this study warrant a further exploration of omission of radiotherapy in selected patients.