We sought to address two Key Questions, each with two sub-questions:

Key Question 1. In adult women with early stage breast cancer, what are the comparative effectiveness, adverse events, and cosmetic outcomes of partial breast irradiation compared to whole breast irradiation?

KQ 1a. How does effectiveness of partial breast irradiation (PBI) vary by clinical-pathologic characteristics?

KQ 1b. How do the effectiveness, adverse events, and cosmetic outcomes of partial breast irradiation vary by target volumes, dose-fractionation schemes, motion management, and planning parameters?

Key Question 2. In adult women with early stage breast cancer, what are the comparative effectiveness, adverse events, and cosmetic outcomes of different partial breast irradiation modalities (including multi-catheter interstitial brachytherapy, single-entry catheter brachytherapy, 3-dimensional conformal external beam radiation therapy, intensity modulated radiation therapy, proton radiation therapy, and intraoperative radiotherapy)?

KQ 2a. When there are no eligible comparative studies to address KQ 2 for a particular PBI modality, what are the rates of adverse events in noncomparative series of such modality?

KQ 2b. When there are no eligible comparative studies to address KQ 2 for a particular PBI modality, what are the rates of long-term (>5 years) effectiveness outcomes and cosmesis in noncomparative series of such modality?

This review also addresses one Contextual Question:

CQ1. In adult women with early stage breast cancer, to what extent does financial toxicity differ between partial and whole breast irradiation?

Findings in Relation to What Is Known

The literature on PBI has proliferated substantially in recent years. However, evidence remains limited about the influence of moderate risk factors in selecting patients most appropriate for PBI treatment, or determining the optimal treatment target volume, radiation dose, fractionation, and modality. Current guidelines recommend PBI as a treatment with similar results to WBI for selected patients. The findings of our systematic review and meta-analyses align with these guidelines and are consistent with results from two recent systematic reviews.

In the era of breast conserving therapy with WBI, IBR was found to be associated with increased risk of distant metastasis and reduced overall survival. In-breast recurrences have been implicated as sources of potential future distant metastases. Therefore, given that we found no difference in IBR between PBI and WBI, it is consistent that we also observed no differences in cancer-free survival or overall survival. Additionally, because WBI exposes more tissue to radiation and can be associated with higher doses to organs at risk, including both the heart and lung, evaluation of overall survival is critically important to evaluate possible effects of treatment-related toxicity outcomes. No statistical differences were identified between overall survival comparing WBI or PBI.

We found that PBI was associated with significantly less acute toxicity compared with WBI. This finding likely reflects the consequences of a smaller amount of irradiated tissue and generally a shorter course of therapy to a lower cumulative dose. In a detailed examination, grade 1 events were not statistically different; however, grade 2 and 3 events were significantly lower. Grade 3 acute toxicity events were very few, but numerically lower for PBI, largely as a result of the differences reported in the GEC-ESTRO study. Just three studies with significant heterogeneity reported on grade 3 events, with wide confidence intervals. The broad applicability of toxicity rates from the GEC-ESTRO study is limited because of the use of the multi-catheter technique, which represents a rarely used treatment option in the United States. Nevertheless, the lack of increased toxicity particularly considering the interventional and technical nature of this modality is noteworthy. Multi-catheter interstitial brachytherapy and IORT were associated with lower rates of acute toxicity compared with WBI. Among studies comparing external beam PBI (intensity-modulated radiation therapy [IMRT] or 3-dimensional conformal external beam radiation therapy [3DCRT]) to WBI, there were heterogenous results with some describing no statistical differences in acute toxicity, and others reporting reduced acute toxicity with PBI. In the very limited data with direct comparison between PBI modalities, both IMRT and single-entry brachytherapy appeared to be associated with lower rates of acute toxicity than 3DCRT. IMRT is associated with improved dose homogeneity, and brachytherapy reduces the volume of breast irradiation, both of which may be associated with reduced acute toxicity.

In review of long-term toxicity, there was no statistical difference between PBI and WBI. With smaller treatment volumes and statistically lower rates of acute AEs, the lack of statistical difference in long-term toxicity is worthy of further exploration. Late AEs in breast cancer patients develop in response to a variety of factors, many of which are relevant to these comparisons, such as normal tissue repair/response to radiation, target volume, dose heterogeneity, and proximity to normal tissues, such as skin or chest wall. Importantly, there is no clear relationship between the risk for acute AEs and the delayed AEs in patients with breast cancer, with some data suggesting that lower risk for acute events does not always translate to lower risk for delayed events. We observed this finding, with lower risk for acute toxicity with PBI, but no difference with late AEs. Delayed events, such as fibrosis, breast lymphedema, telangiectasia, and fat necrosis are mediated by different, but related pathways.98 These events may also be more sensitive to dose and fractionation, as well as normal tissue recovery, such as sub-lethal damage repair pathways. Similarly, the risks for developing late AEs from breast radiotherapy of any kind may be reduced over the last two decades through technological improvements in the target definition, localization, and dose distributions. Improved dose distributions in breast radiotherapy have showed lower rates of AEs, and this has been observed in comparisons of PBI with 3DCRT and IMRT. 73 This highlights the challenge of evaluating outcomes over time among treatments with evolving technology. For example, while some institutional reports of early accelerated partial breast irradiation (APBI) experiences reported high rates of cosmetic deterioration, in the most recent published RCTs, the risks for both cosmetic change and AEs were lower with ABPI. Similarly, early experiences with PBI using passive scatter proton therapy report high rates of late skin toxicities, which have not been observed in more recent studies with updated treatment and planning techniques. For example, telangiectasia of any grade is reported in as many as 69 percent of patients treated with passive scatter proton therapy, in contrast to as few as 1.3 percent with pencil beam scanning.

The lack of significant difference in cancer-free survival and overall survival between PBI and WBI is consistent with other systematic reviews. One meta-analysis reported a decrease in nonbreast cancer mortality with PBI compared with WBI; however, there are methodological concerns regarding the selection of studies and reported results. While the TARGIT-A trial reported reduced nonbreast cancer deaths favoring IORT compared with WBI, this observation is inconsistent with expectations for the timeline to development of AEs and unsupported by other evidence examining IORT, as well as numerous other studies of PBI that also had excellent cardiac sparing.

Implications for Clinical Practice

Appropriate patient selection is a critically important aspect of the success of PBI. There is broad consensus in multiple treatment guidelines and systematic reviews that PBI is an acceptable treatment option for patients with clinical and tumor characteristics similar to those represented on clinical trials, for example, postmenopausal age range, estrogen receptor (ER) positive status, grade 1-2, no lymph node involvement, and tumor size ≤2 cm. The results presented in this report represent data from 15,276 patients who participated in RCTs of PBI versus WBI, more than three-fold the number of patients who participated in clinical trials that led to the adoption of breast conserving surgery and WBI as a standard treatment approach. In aggregate, the results of our meta-analysis and systematic review showed no difference between PBI and WBI for selected patients. The finding of reduced acute toxicity with PBI represents a significant finding that will meaningfully inform patient and physician decision making.

Uncertainty remains regarding the magnitude of increased risk associated with features that are perceived as less favorable that were included within the eligibility criteria but represent a minority of patients who participated, for example age <50 years, invasive lobular carcinoma, tumor size 2.1-3 cm, grade 3, ER negative status, Human Epidermal Growth Factor Receptor 2 (HER2) positive status, positive for lymphovascular invasion, or elevated Ki-67. Our analysis revealed the lack of data on the outcomes in these subgroups and highlight the importance of future investigation to develop more robust evidence to inform treatment recommendations.

This review does not provide cost information.

Limitations and Future Research

The clinical trials included in our aggregate analysis represent a variety of treatment techniques, including several methods of external beam radiotherapy (3DCRT, IMRT, proton therapy), brachytherapy (multi-catheter interstitial brachytherapy, single lumen applicator brachytherapy, multi-lumen applicator brachytherapy) and IORT (low-energy x-ray, electrons). Treatment outcomes of each individual radiation modality were insufficiently reported, which limited the ability to make comparisons between modalities.

Evaluation of outcomes according to patient, tumor and treatment subgroups was similarly limited by the available data. Many of the included clinical trials did not report subgroup analyses, and often the subgroups were not able to be combined for aggregate analysis. As a result, we were unable to assess many of the prespecified subgroups. Additionally, the results of subgroup analysis are limited by sample size and the risk of false-positive or false-negative findings. Our results from subgroup analysis may be informative in directing future areas of investigation but cannot definitively determine the magnitude of risk associated with each characteristic. Our results highlight the importance of further investigation to determine the outcomes of PBI among patients with adverse risk factors.

Although evaluation of cosmetic outcome was consistently reported using a 4-point scale to describe excellent, good, fair, and poor cosmesis (Table 2) in a relatively homogeneous population of favorable prognosis breast cancer, the reported rate of provider-assessed fair or poor cosmesis for WBI versus PBI spans a wide range. For example, three studies reported significantly higher rate of fair or poor cosmesis with WBI, while another study reported significantly higher rate of fair-poor cosmesis with PBI, with a spectrum of results between the two extremes. This variability between studies is reflected in the finding of statistically significant heterogeneity on 5-year and 10-year followup results for both provider-reported (I²=89% and I²=94%, respectively) and patient-reported cosmesis (I²=56% and I²⁼96%). Substantial heterogeneity, risk of bias and imprecision across studies, and lack of cosmetic outcome data from the largest study of PBI (NSABP B-39) limits the ability to draw conclusions regarding cosmetic outcomes in a comparison of PBI versus WBI.

We note that increased use of oncoplastic surgery in addition to lumpectomy may influence both eligibility for partial breast radiotherapy as well as the cosmetic outcome. In the reported studies of partial breast radiotherapy, oncoplastic tissue rearrangement was not specified as an exclusion criterion (with one exception)40 but was not commonly used when the studies of PBI were developed. It is widely accepted that oncoplastic surgery typically precludes the ability to define the lumpectomy cavity for partial breast radiotherapy. The use of IORT for a boost after lumpectomy has been described, and has been used as a method of delivering the boost in the setting of oncoplastic tissue rearrangement.

Several outcomes represent provider-rated or patient-reported measures, such as cosmesis, AEs, and quality of life. The comparison of PBI and WBI is not blinded to either clinicians or patients, and it is possible that the treatment assignment might have influenced perceptions of these subjective measures.

We could not statistically evaluate publication bias in all of the comparisons because the number of studies included in these comparisons was small (n<10). Only studies published in English language were included in this review.

There is a critical need to further evaluate patient reported outcomes and their influence on decision-making for breast radiotherapy. Similar oncologic outcomes between modalities, as reported here, suggest that key therapeutic differentiators for patients often lie within the expected toxicity and quality of life. Some patients experiencing cancer are willing to sacrifice some efficacy of therapy for maintenance of quality of life, with substantial variation depending on factors such as age and baseline health status. Indeed, patients with favorable risk DCIS or older patients with early stage breast cancer may opt for omission of radiotherapy altogether, thus limiting the potential impact of treatment on quality of life and financial toxicity, notwithstanding an increased but accepted risk of IBR. Patient valuation of benefits and risks of radiotherapy represent an important area for future study, particularly as radiotherapy omission has become a key area of ongoing investigation among younger patients with biologically favorable tumors, such as those enrolling on the DEBRA trial (NRG BR007). Early results of the LUMINA trial support the hypothesis that highly selected women with Luminal A-like tumors might have a sufficiently low rate of local recurrence to forego radiotherapy, provided that there is compliance with a complete course of endocrine therapy. Older women with favorable-risk breast cancer may prefer to avoid side effects related to endocrine therapy, and often choose a single treatment modality with either exclusive endocrine therapy or exclusive PBI alone. Comparison of these two approaches is currently being evaluated on the EUROPA clinical trial. The landscape of options for early-stage favorable risk breast cancer is a key area of ongoing investigation that will significantly influence future decisions about tailoring the use of radiotherapy.

Additionally, with results from the FAST-Forward clinical trial showing non-inferior breast cancer outcomes and similar adverse effects with "ultrahypofractionated" WBI completed in 5 fractions compared to conventionally hypofractionated WBI,10 the use of accelerated WBI is becoming more widely adopted in practice. The availability of "ultrahypofractionated" WBI narrows the distinction between PBI and WBI, since both can be completed within 1 week, with similar side effects and reduction in the financial burden of treatment. It is plausible that patients considered as "cautionary" or "unsuitable" for partial breast radiotherapy according to ASTRO criteria might consider accelerated WBI rather than PBI. Notwithstanding this alternative, many patients may be motivated to pursue PBI to minimize radiation exposure of the breast and adjacent normal tissue, and thus defining the suitability of PBI in moderate risk subgroups remains an area of interest for future study.

Finally, radiotherapy technology has developed and dramatically changed over the past two to three decades, with a transition from 2D radiotherapy to routine use of 3D radiotherapy, IMRT, and other advanced planning technologies. These advancements result in improved dose homogeneity with fewer “hot spots,” which may lower the risk for adverse events. In addition, localization with image guidance enables smaller PTV expansions and improved treatment accuracy, which introduces challenges in comparisons that span a wide time interval of significant changes in radiotherapy technology and treatment. Although the volume of the treatment target relative to the breast, dose/fractionation schedule, and planning parameters are recognized as critically important to understand the risks related to treatment, these data were very limited or unavailable for many studies. Defining an optimal radiation dose, fractionation, and target size using contemporary techniques for treatment planning and image guidance, and characterizing the outcomes of that approach, represent key areas for future study, particularly for short regimens with daily treatment.