Patterns of Post-Mastectomy Radiotherapy in Immediate Breast Reconstruction – Results from the iBRA-2 Cohort Study

Purpose Long-term data indicates that post-mastectomy radiotherapy (PMRT) is associated with improved overall survival in all node-positive breast cancer patients. Immediate breast reconstruction (IBR) remains controversial in the context of planned PMRT, but rates of IBR are increasing. The aim of this study was to examine current practice of PMRT in patients undergoing mastectomy +/- IBR. Methods Data were collected from 2,526 patients enrolled in the iBRA-2 prospective cohort study undergoing mastectomy +/- IBR between 1 st July and 31 st December 2016, recruited consecutively at 71 centres across the United Kingdom, Ireland, and ve international centres.


Introduction
Breast cancer is the most common female cancer in Europe with 562,000 new cases diagnosed in 2018 [1]. Despite advances in treatment and early diagnosis, almost 40% of women undergo mastectomy as their primary surgery [2,3]. After surgery, radiotherapy is the second most commonly used treatment for breast cancer. Long-term data combining a number of previous randomised-controlled trials demonstrates that post-mastectomy radiotherapy (PMRT) reduces recurrence rate and improves survival in all node-positive breast cancer patients [4]. As a result, the indications for PMRT in women with breast cancer are widening. Population-based data from the US shows that use of PMRT in patients with 1-3 positive lymph nodes increased signi cantly between 2003 and 2012 [5,6]. Several national treatment guidelines now recommend that radiotherapy be considered for all patients with node-positive disease after mastectomy [7][8][9].
Women undergoing mastectomy are offered immediate breast reconstruction (IBR) with the aim of improving quality of life [10]. IBR remains controversial in the context of planned PMRT [11]. However, an analysis of the US Surveillance, Epidemiology, and End Results (SEER) database indicates that an increase in PMRT amongst women with stage I to III breast cancer undergoing mastectomy did not lead to a concomitant decrease in IBR [6]. In fact, data from the US Mastectomy Reconstruction Outcomes Consortium (MROC) for >2,000 patients demonstrates that around 25% of implant-based and half of autologous reconstructions received PMRT [12]. The UK National Institute of Health and Care Excellence (NICE) recently reviewed 23 observational studies and concluded that there was insu cient evidence to indicate worse patient outcomes from PMRT in the setting of IBR, and that IBR should be offered even if adjuvant RT is anticipated [13].
Currently, high quality evidence for the use of PMRT in intermediate-risk breast cancer (de ned as pT1-2N1; pT3N0; or pT2N0 if also grade III or with lympho-vascular invasion) is lacking and the results of a large European randomized-controlled trial are awaited [14]. At the same time, rates of IBR in the UK, in particular implant-based reconstruction, are increasing [15], yet large scale population-based data on the use of PMRT in patients undergoing IBR has not been published to date. In reconstructive breast surgery, the trainee research collaborative model has recently emerged as a time and cost-effective method for delivering large-scale prospective studies [16,17].
This Breast Reconstruction Research Collaborative of breast and plastic surgeons was utilized to deliver the iBRA-2 study to determine the impact of IBR on the delivery of adjuvant treatment [18]. The present study was undertaken to analyse patterns of PMRT amongst patients undergoing mastectomy with and without IBR enrolled in the iBRA-2 cohort.

Methods
The methods of the iBRA-2 cohort study have been reported elsewhere [18,19] Consecutive women aged 18 or over undergoing mastectomy with or without IBR using any technique for invasive or pre-invasive (ductal carcinoma in situ, DCIS) breast cancer with curative intent were included. Patients were identi ed from multidisciplinary team (MDT) meetings; unit operating records and clinics.
Excluded were patients undergoing risk-reducing surgery (without a therapeutic mastectomy for breast cancer), partial mastectomy including wide local excision with volume replacement (latissimus dorsi mini-aps; lateral intercostal perforator (LICAP) or thoracodorsal artery perforator (TDAP) aps) or displacement techniques (therapeutic mammaplasty), and those with distant metastatic disease.
The iBRA-2 study was classi ed as service evaluation according to the NHS Health Research Authority online decision tool http://www.hra-decisiontools.org.uk/research/, so ethical approval was not required.
Each participating centre obtained local governance approvals prior to entering patients into the study. Data were collected prospectively, including baseline demographic and operative data. Oncological data and adjuvant treatment recommendations were collected from post-operative MDT meeting records. Pilot data collected at a number of participating sites between 1st May and 30th June 2016 suggested that adjuvant therapy was unlikely to commence earlier than six weeks post-operatively. Data collection in patients not requiring adjuvant treatment therefore continued from the last de nitive cancer surgery until six weeks following surgery.
Study data were collected and managed using REDCap (Research Electronic Data Capture) electronic data capture tools hosted at the Kennedy Institute of Rheumatology, University of Oxford [20,21]. REDCap is a secure, web-based software platform designed to support data capture for research studies, providing 1) an intuitive interface for validated data capture; 2) audit trails for tracking data manipulation and export procedures; 3) automated export procedures for seamless data downloads to common statistical packages; and 4) procedures for data integration and interoperability with external sources. For quality assurance purposes, the lead investigator at each site was asked to independently validate 5-10% of the data. If concordance between the data entered on REDCap and that independently-validated was <90%, the unit's data were excluded from the analysis, consistent with the QA procedure used in other collaborative projects.

Study de nitions and endpoints
Primary and secondary outcomes in iBRA-2 were selected based on current best practice [22] and national breast cancer guidelines [23]. The endpoint in the present study was use of adjuvant radiotherapy, de ned as the recorded post-operative MDT outcome "radiotherapy recommended" or "radiotherapy to be discussed". For analysis purposes, patients in iBRA-2 were categorised into four groups according to the most complex surgical procedure received as: i) mastectomy only (no reconstruction), ii) mastectomy and IBR with implant-only techniques, iii) mastectomy and IBR with pedicled aps, and iv) mastectomy and IBR with free-ap techniques.
Implant-based procedures included any reconstruction in which only expanders/implants were used to reconstruct the breast, either with or without biological (e.g. acellular dermal matrix) or synthetic (e.g titanium-coated polypropylene) mesh irrespective of whether the implant/expander was placed in a pre or sub-pectoral position. Pedicled-ap procedures included any pedicled-ap used to reconstruct the breast with or without an implant/expander, including latissimus dorsi (LD) and transverse rectus abdominus myocutaneous (TRAM) aps. Free-ap procedures included any technique in which a microvascular freeap was used for IBR, including deep inferior epigastric perforator (DIEP), super cial inferior epigastric perforator (SIEA), superior and inferior gluteal artery perforator (SGAP and IGAP), and transverse upper gracilis (TUG) aps. All complications were de ned a priori [19]. Major complications were de ned as any complication requiring re-admission or re-operation. Minor complications were de ned as those managed conservatively. An analysis of short-term complications is described elsewhere [18].

Statistical analysis
Descriptive summary statistics were calculated for each variable in the whole cohort and sub-divided by procedure type (mastectomy only; implant-based reconstruction; pedicled reconstruction; and free-ap reconstruction). Categorical data were summarised by counts and percentages, and continuous data by median, interquartile range (IQR), and range. Procedure groups were compared using appropriate nonparametric statistics. Univariable and multivariable logistic regression were used to explore associations between clinico-pathological variables and use of PMRT, including patient and procedure-related factors, namely, age, body mass index (BMI), smoking status, American Society of Anaesthesiologists' (ASA) grade; diabetes, ischaemic heart disease (IHD); other comorbidities, chemotherapy, ER-and HER-2 status, lymphovascular invasion, tumour grade, multi-focality, and procedure type.
To assess selection bias, correlation between proportions of patients undergoing PMRT and IBR was determined by region using Pearson's r. Regional trends in PMRT across the United Kingdom and Ireland were further assessed in multivariable logistic regression, adjusted for BMI, co-morbidities, smoking status, ASA, ER-and HER2-status, lymphovascular invasion, multi-focality and tumour grade, and including clustering by centre. All analyses were performed in STATA 15 (STATA Inc, Texas).

Results
Of 2,526 patients recruited into the iBRA-2 study, 666 underwent implant-based reconstruction, 105 received pedicled aps, and 227 underwent free-ap reconstruction. The remainder of patients underwent mastectomy only without reconstruction. As 80 patients had bilateral cancer surgery and there were 16 procedures with missing outcome data, a total of 2,590 breast procedures were included in the analysis. On a per breast basis, Table 1 summarises patient demographics and adjuvant therapy decisions by procedure type. Patients undergoing mastectomy alone (n=1,559 breast procedures) were older than those undergoing IBR and were more likely to have co-morbidities as indicated by ASA grade. Patients undergoing IBR were more likely to have lower stage cancers. Adjuvant chemotherapy was more likely to be recommended after mastectomy-only than after IBR. Adjuvant radiotherapy was recommended or for discussion in 44.9% of mastectomies, 31.4% of implant-based reconstructions, and in 34.3% and 32.5% of pedicled and free-ap reconstructions, respectively (p<0.001). PMRT recommendation by tumour and nodal stage Figure 1 shows the proportion of patients on a per-breast basis (n=2,590) recommended for PMRT by tumour stage. Out of all patients with T3, N2 or N3 stage disease, 89% were recommended for radiotherapy, while 64.7% of patients with N1 stage disease were recommended for radiotherapy. Interestingly, 13.2% of Tis and 21.1% of N0 stage patients were also recommended for radiotherapy. When sub-divided by tumour and nodal stage, there were no differences between surgical procedure types (see Supplementary Table).

Impact of PMRT use on IBR
Excluding patients from ve centres outside the United Kingdom and Ireland, Figure 2 shows no clear correlation between the proportion of patients recommended for PMRT and of the proportion receiving IBR across the 14 regions (r=-0.23, p=0.41).

Discussion
The aim of this study was to analyse patterns of PMRT amongst patients undergoing mastectomy with and without IBR. The short-term outcomes and rates of complications in the iBRA-2 patient cohort have been reported elsewhere [18]. In relation to PMRT, the present study found that between 31.4% and 34.3% of women undergoing IBR were recommended for or for discussion of radiotherapy compared to 44.9% of those receiving simple mastectomy. These gures are in keeping with several previously published observational studies showing that between 29% and 43% of patients with IBR are irradiated [12,17,24,25]. Although IBR was more likely to be performed for lower stage cancers and in younger patients with fewer co-morbidities, women undergoing IBR were no less likely to be recommended for radiotherapy after adjusting for patient and tumour variables. This adds to a growing body of evidence that in the setting of PMRT, IBR is becoming an acceptable option for women requiring PMRT [26-28] and that PMRT is only one of many factors in the decision for/against IBR which include consideration of patient and oncological factors as well as surgeon and patient preferences. Indeed, it has been shown that breast cancer multi-disciplinary teams (MDTs) cannot predict with certainty whether a patient will require adjuvant PMRT [29].
We found no clear correlation between higher proportions of patients recommended for PMRT and lower proportions of IBR across different UK regions. This suggests that participating centres offered all types of IBR irrespective of whether PMRT was likely to be needed in the adjuvant setting, in keeping with ndings from a similar US population-based study [6] and current clinical guidelines [13]. The individual choice of IBR technique is likely to depend on several patient and surgical factors as well as locally available expertise and access to immediate free-ap reconstruction. However, in patients requiring PMRT, implant-only IBR is generally associated with more complications compared to autologous IBR [12,17]. Early follow-up data from two prospective cohort studies shows lower satisfaction among patients with irradiated implant-only reconstruction [30,31]. Nevertheless, further work is required to establish the impact of PMRT on the long-term outcomes of IBR to provide high-quality data to inform practice. Initial results from non-randomised studies comparing neoadjuvant versus adjuvant radiotherapy in patients scheduled for IBR are promising and demonstrate the feasibility and safety of this approach [32] but the results of randomized trials will take some years to mature.
Compared to previous population-based analyses showing between 30.3% and 40.5% of patients with N1 disease (1-3 positive lymph nodes) receiving PMRT by 2012 [5,6], in our study, 64.7% of such patients were recommended for PMRT. This con rms the trend of an increasing number of patients with N1 disease being considered for PMRT and it is in keeping with national clinical guidelines [7][8][9].
Furthermore, the number of patients in our study with non-invasive cancer receiving post-mastectomy radiotherapy appears relatively high compared to an earlier UK study [33], although our study was not restricted to the breast screening population, and the receipt of radiotherapy following mastectomy y for DCIS may be related to margin status, grade, and presence of micro-invasion.
Subdividing the data by UK regions, we found in our study that patients in Northern Ireland/Ireland, Scotland, the North of England and the West Midlands were signi cantly less likely to be recommended for PMRT. This highlights some interesting geographical variation, which merits further investigation. A similar US-based study found that density of radiation oncology practices was associated with receipt of PMRT in patients with N1 stage breast cancer [5]. Results of the Selective Use of Post Mastectomy Radiotherapy (SUPREMO) trial are awaited and are likely to reduce the variation in practice of PMRT use in intermediate-risk breast cancer (de ned as pT1-2N1; pT3N0; or pT2N0 if also grade III or with lymphovascular invasion) [14].
There is accumulating evidence that individual patient and tumour molecular pro les will be able to predict the response to adjuvant treatment. Several phase III trials are currently underway [34] or have already reported on the effectiveness of the 70-or 21-gene recurrence scores in selecting patients for adjuvant chemotherapy [35,36]. Genomic-adjusted radiation dose (GARD) therapy may allow personalisation of radiotherapy on the basis of the biological effect of a given physical dose of radiation, calculated using individual tumour genomics, and has been shown to predict time to recurrence and overall survival [37]. In terms of complications from IBR, predictive risk models have been externally validated with receipt of PMRT (or not) as a predictor [38,39]. Other risk prediction models for individual radiotherapy side-effects in the breast have been validated [40,41], though none in the setting of IBR. In the future, personalised medicine approaches including the patient germline and tumour genomic pro le may aid decision-making around IBR and PMRT.

Strengths and Limitations
To our knowledge, this is the largest prospective cohort study to date in the UK and Ireland investigating patterns of PMRT in patients undergoing mastectomy with and without IBR, but there are several limitations. Due to its observational design, the study is at risk of bias. Patients were recruited consecutively from participating centres but there were baseline differences between the treatment groups. Although we adjusted for known confounders such as cancer stage, patient age, BMI, smoking status, and ASA grade, we acknowledge that it is not possible to adjust for all potential confounders. Our study included patients from 71 centres in the UK and Ireland as well ve centres in Italy in Egypt. It is possible that participating units differed from those not taking part. However, approximately half of all UK breast units were represented in the study.
Due to the short follow-up period (up to initiation of rst adjuvant treatment), we were only able to analyse treatment recommendations and were unable to ascertain successful receipt or completion of PMRT in all patients. It was not possible to follow patients up to completion of adjuvant therapy with this surgical trainee collaborative design, but collaborations with oncology trainees should allow us to address this issue in the future. Moreover, a data-linkage study is planned in future to explore oncological outcomes in terms of local recurrence, disease-free and overall survival. While it is not possible to establish causation with this observational study design, the design and conduct of randomized controlled trials in the setting of IBR are perceived as di cult due to patient and surgeon's preference for type and timing of breast reconstruction [25].

Conclusions
The present study provides evidence that approximately one third of patients undergoing IBR were recommended for PMRT. About two thirds of patients with T1-2 N1 disease were recommended for PMRT, which re ects current clinical treatment guidelines for this patient group. Within the UK and Ireland, the study also highlighted regional variation in PMRT practice which merits further investigation. Although IBR was more likely to be performed for lower stage cancers and in younger patients with fewer comorbidities, patients undergoing IBR were no less likely to be recommended for radiotherapy, irrespective of reconstruction technique. Further work is required to establish the impact of PMRT on the long-term outcomes of different types of IBR to help women and surgeons make more informed decisions about breast reconstruction options.

Figure 1
Proportion of patients (%, per-breast data) in the study recommended for post-mastectomy radiotherapy by tumour stage.