The Efficacy of Low-Kilovoltage X-Rays Intraoperative Radiation as Boost for Breast Cancer: A Systematic Review and Meta-Analysis

Background Intraoperative radiotherapy (IORT) is a novel promising technology that may replace external beam radiation therapy (EBRT) as boost for patients receiving breast-conserving surgery. To better evaluate the efficacy of IORT using low-kilovoltage (low-kV) X-rays as boost, we presented this meta-analysis according to the PRISMA checklist. Methods Studies reported survival outcomes of intraoperative radiation using low-kilovoltage X-rays system (Intrabeam®, Carl Zeiss Meditec, Dublin, CA, USA) as boost were identified through electronic bibliographic database: PUBMED. The meta-analysis module in Stata (16.0) is used to pool the studies. A Poisson regression model is used to predict a 5-year local recurrence rate. Results Twelve studies including 3006 cases were included in the final analysis, with a median follow-up of 55 months weighted by sample size. The pooled local recurrence rate is 0.39% per person-year (95% CI: 0.15%–0.71%), with a low degree of heterogeneity (I2 = 0%). The predicted 5-year local recurrence rate was 3.45%. No difference in pooled local recurrence rate was found between non-neoadjuvant patients studies and neoadjuvant patients studies (0.41% per person-year vs. 0.58% per person-year, P = 0.580). Conclusions This study shows that low-kV IORT is an effective method as boost in breast cancer patients, with a low pooled local recurrence rate and low predicted 5-year local recurrence rate. Besides, no difference in the local recurrence rate was found between non-neoadjuvant patients studies and neoadjuvant patients studies. Low-kV IORT boost may be a promising alternative to EBRT boost in the future, which is being tested in the ongoing TARGIT-B trial.


Introduction
Worldwide, breast cancer has been the most common carcinoma in the women population [1]. Te treatments of breast cancer mainly include locoregional treatment and systematic treatment. Breast-conserving surgery (BCS) combined with radiotherapy has been proven to be an effective locoregional treatment and widely accepted since 1985 [2,3]. In clinical practice, whole breast irradiation with or without boost is the standard radiotherapy treatment after BCS. A phase III randomized trial indicated that the boost group has a lower 20-year cumulative incidence of ipsilateral breast tumor recurrence than the no-boost group (12.0% vs. 16.4%) [4]. Traditionally, external beam radiation therapy (EBRT) was used to deliver boost dose to the tumor bed, taking 5-7 days generally.
Recently, intraoperative radiotherapy (IORT) emerged as an optional method for tumor bed boost which can be performed concurrently with surgery. With an applicator placed in the tumor bed after lumpectomy, IORT can deliver the prescribed dose to the breast tissue. Te Intrabeam ® system (Carl Zeiss Meditec, Dublin, CA, USA) is one of the IORT equipment that uses low-kilovoltage (low-kV) X-rays. TARGIT-A (NCT00983684) was a large randomized, noninferiority trial that proved low-kV X-ray IORT was an efective alternative to EBRT after BCS, with comparable long-term efcacy for cancer control. At 12-year follow-up, the nonbreast cancer mortality was signifcantly lower with low-kV IORT (5.41% vs. 9.85%, HR = 0.59, P � 0.005), mainly due to fewer deaths from cardiovascular disease, lung problems, and other cancers [5,6].
Te initial series of patients treated with low-kV IORT as an intraoperative boost suggested that it might provide superior local control rates for BCS [7]. Te TARGIT-B randomized clinical trial [8], currently recruiting in 38 centers, is comparing low-kV IORT boost with EBRT boost. Tis trial is testing whether low-kV IORT boost is superior to EBRT boost in terms of local control and survival.
Tis study aims to pool the low-kV IORT boost studies into a meta-analysis, enriching the population of the sample so as to assess the efcacy of low-kV IORT as boost. We presented this article in accordance with the PRISMA reporting checklist (Supplementary fle).

Evidence Acquisition.
A prospective protocol of objectives, literature-search strategies, inclusion and exclusion criteria, outcome measurements, and methods of statistical analysis was prepared according to the Preferred Reporting Items for Systematic Reviews and Meta-analysis and Metaanalysis of Observational Studies in Epidemiology recommendations for study reporting [9,10]. Neither the review nor protocol was registered before. Studies had to meet two criteria for inclusion. Tey should have investigated recurrence data of breast cancer patients and have used low-kilovoltage X-rays system (Intrabeam ® , Carl Zeiss Meditec, Dublin, CA, USA) as intraoperative boost radiation. To sum up, reviews, case reports, and studies use other IORT systems, and studies not using IORT as boost, and studies without recurrence data. Besides, studies not included by Jounral Citation Reports (JCR) were all excluded. To further supplement the database searches, a review of each included study was completed. When multiple reports describing the same population were published, the most recent or complete report was used. Two reviewers (Yuanjian Fan and Zhen Shan) screened each record, and each report was retrieved independently.

Data Extraction. Two reviewers (Yuanjian Fan and
Zhen Shan) extracted the data and summarized it independently. Any disagreement was resolved by the adjudicating senior authors. Since the included studies are all nonrandomized, we used the Newcastle-Ottawa Scale (NOS) to assess the quality of all studies, which consists of three parts: selection, comparability, and outcome [11].
In this meta-analysis, the main outcome of interest was the local recurrence rate (LRR). We further investigated the included studies and defned any local recurrence as events. Data were extracted from each of the included studies regarding the characteristics related to the study, protocol, and patients.

Statistical Analysis.
We used meta-analysis to provide a pooled summary of the data on the local recurrence rate. To obtain a pooled estimate rate of "events," we used the Metaprop module in Stata (Version 16.0). Te original pooled estimate LRR was counted in per person-year, which means LRR for every single patient in every single year. A randomized-efect meta-analysis of proportion models was used to estimate an overall LRR. Of note, we specifcally calculate the pooled LRR within two subgroups: non-NAT (neoadjuvant treatment) patients studies and NAT patients studies, to further assess the efcacy of IORT as boost in diferent subgroups.
Because the majority of the studies did not follow up for over 5 years, it was difcult to estimate a reliable 5-year LRR. Terefore, Poisson regression modeling for the pooled recurrence rate was used to estimate a reliable 5-year LRR [12].

Heterogeneity.
We estimated the heterogeneity between studies with the I 2 statistic, which described the percentage of variation between studies due to heterogeneity rather than chance. Te values of 25%, 50%, and 75% show low, moderate, and high degrees of heterogeneity, respectively [13]. To draw a conclusion that can be extrapolated to more breast cancer patients, we used a random-efects metaanalysis proportions model to calculate the LRR.

Included Studies and Patients' Characteristics.
Eventually, 12 studies including 3006 cases were included in the fnal analysis, with a median follow-up of 55 months weighted by sample size (Table 1 and Tables 2 and 3.

Discussion
Te advantages of using low-kV IORT include the ability to visualize the tumor bed directly. Surgeons can deliver a single dose of radiation to the surrounding tissue intraoperatively, ensuring the treatment of the high-risk tissue and eliminating the risk of marginal missing. Patients who undergo IORT boost can omit postoperative EBRT boost which may cost 5-7 days generally. Besides, the cosmetic and toxicity outcomes of low-kV IORT boost are good because of the lower doses [15,25]. In the TARIGIT-A trial, there was no signifcant diference in any protocoldefned wound-related complications such as fbrosis, breast edema, retraction, ulceration, lymphedema, hyperpigmentation, and pain. Fewer grade 3 or 4 radiotherapyrelated skin complications are associated with low-kV IORT patients than with EBRT (4/1721 vs. 13/1730, P � 0.029) [5].
Both the pooled LRR (0.41%) and the predicted 5-year LRR (3.45%) are relatively low in overall patients. When NAT patients' studies were excluded, we may achieve a relatively lower predicted 5-year LRR than overall studies (2.66% vs. 3.45%). Te previous study reported the 5-year LRR of 4.3% (95% CI: 3.8%-4.7%) in patients who received BCS plus EBRT boost [26], which seems to be higher than that of IORT as boost in our study. However,   such diferences may refer to the improvement of adjuvant therapy.
A meta-analysis carried by Early Breast Cancer Trialists' Collaborative Group (EBCTCG) indicates that tumors downsized by NAT might be associated with a higher local recurrence risk after BCS, comparing to tumors of the same dimensions in patients who received adjuvant chemotherapy instead [27]. Multiple reasons may contribute to the higher local recurrence rate for NAT patients. Firstly, NAT is usually prescribed to those patients with high risks, such as large tumor size, high tumor grade, lymph node involved, HER2 positive, or triple-negative disease. Tese features make the prognosis worse than that of non-NAT patients. Secondly, it is difcult to localize the primary tumor bed precisely after NAT, especially for tumor with good response. Tis may lead to the high risk of missing tumor bed irradiation. In our study, the local recurrence rate of non-NAT patients studies and NAT patients studies showed no signifcant diference (0.41% per person-year vs. 0.58% per person-year, P � 0.580). However, it is a far cry from the LRR of two included non-NAT patients' studies (2.81% per person-year vs. 0% per person-year). We supposed that such diference mainly refers to the patients' characteristics. Although the exact NAT patients' characteristics are unavailable in Cho  showed that low-kV IORT boost is superior to EBRT boost in NAT patients (local recurrence-free survival 88.5% vs. 79.9%). Te efcacy of low-kV IORT boost in NAT patients needs more validation, as is being done in the TARGIT-B randomized trial (NCT01792726) [8].
Intraoperative electron radiotherapy (IOERT) as boost has been proved to be an efcacy radiotherapy prior to WBI, with outstanding local control rates. In a long-term result of a phase III randomized study included 133 patients using IOERT as boost, only 0.8% of 5-year in-breast true recurrences was observed [28]. A large pooled analysis compared 1109 unselected patients from 7 diferent centers using the same IOERT and WBI doses: 10 Gy as a boost and 50-54 Gy WBI. At a median follow-up of 72.4 months, 99.2% of the tumor control rate was achieved [29]. In our study, only 4 of the included studies reach a median follow 35% (4/170), and 7.35% (5/68), respectively. All of them are higher than that of IOERT as boost. Head-to-head studies comparing low-kV IORT boost, EBRT boost, and IOERT boost are necessary to further compare the efcacy of multiple methods of boost.

Limitations
Tere are several limitations in our study. Firstly, all the included studies are nonrandomized studies, leading to unavoidable selection bias. Besides, all studies were carried out in diferent clinical centers, which may result in diferent IORT protocols. Tirdly, most of these studies (10 of 12) are single-armed studies and have no control group, which make it difcult to compare low-kV IORT with other boost methods.

Conclusion
Tis study shows that low-kV IORT is an efective method as boost in breast cancer patients, with a low pooled local recurrence rate and a low predicted 5-year local recurrence rate. Besides, no diference of the local recurrence rate was found between non-neoadjuvant patients' studies and neoadjuvant patients' studies. Low-kV IORT boost may be a promising alternative to EBRT boost in the future, which is being tested in the ongoing TARGIT-B trial.

Data Availability
All the data in our study can be accessed from the 12 included studies of meta-analysis.