Multiple Primary Cancer and Survival After Lumpectomy with or without Radiotherapy for Ductal Carcinoma in Situ: A Propensity-Score-Matching Study

Background: Radiotherapy combined with breast-conserving surgery is widely performed in patients with ductal carcinoma in situ (DCIS). This research was conducted to evaluate the use of radiotherapy to reduce the risk of multiple primary cancer (MPC) and mortality in DCIS patients. Methods: 108,416 patients first diagnosed with DCIS between 1998 and 2015 who received lumpectomy in the Surveillance, Epidemiology, and End Results (SEER) 18 database were included. Age, race, year of diagnosis, laterality, pathologic grade, surgery, radiation, estrogen receptor status, progesterone receptor status, tumor size and vital status were extracted. A comparison of lumpectomy vs. lumpectomy plus radiotherapy was performed using 1:1 propensity score-based matching. Results: Of the 108,416 patients, 39,039 patients were treated with lumpectomy alone, and 69,377 were treated with lumpectomy and radiotherapy. The adjusted hazard ratio (HR) for death was 0.801 and 0.7444 for occurrence of MPC in the lumpectomy and radiotherapy vs. lumpectomy alone groups, respectively. Conclusion: Lumpectomy plus radiotherapy is associated with a significant reduction in mortality and risk of MPC, mainly second primary BC. Younger patients, Black women with high-grade tumors were likely to benefit most. Radiotherapy reduced the risk of occurrence rather than the mortality of MPC patients to reduce the overall mortality.


Introduction
Ductal carcinoma in situ (DCIS) is a noninvasive condition in which abnormal cells are found in the lining of a breast duct 1 .
Since the 1970s, multicenter prospective randomized controlled studies on breast-conserving surgery for early breast cancer (BC) patients have shown no significant difference in disease-free and overall survival in such patients compared to those receiving total mastectomy 2,3 . In addition, radiotherapy plays an essential role in breast-conserving surgery 4,5 , resulting in low mortality for patients with DCIS 6,7 . While long-term survivors benefit from systemic treatment, the frequency of multiple primary cancer (MPC) diagnosis after BC has been increasing, with MPC becoming one of the leading causes of death 8 .
Multiple primary cancer (MPC) is defined as more than 2 separate original tumors that simultaneously or successively occur in the same or different organs 9 . Previous studies reported a greater risk of MPC among patients with an initial diagnosis of breast cancer at the age of <50 years and in Black women with an ER-/PR-status than in Other patients 10 . A retrospective study did not demonstrate an association between systemic therapy and increased risks of MPC`1 1 , while a nested case-control study using the SEER database suggested that radiotherapy (HR=1.33) was related to the risk of second primary malignancy after BC 12 .
To further study the risk of MPC and mortality in DCIS patients, we conducted a population-based study of women with DCIS using the Surveillance, Epidemiology, and End Results (SEER) database. We extracted data on age and year of diagnosis, laterality, pathologic grade, treatment (surgery and radiation), estrogen receptor (ER)status, progesterone receptor (PR) status, tumor size, MPC and death from BC. We sought to determine whether radiotherapy was associated with a reduced risk of BC death and MPC in this cohort of women and to identify subgroups of women who might benefit the most from radiotherapy.

Study population
SEER 18 Regs Custom Data (with additional treatment fields), Nov 2017 Sub (1973-2015 varying) was used to identify the cohort of women for this study. Since estrogen receptor status and progesterone receptor status were provided after 1998, we selected all female patients with confirmed stage 0 BC coded by the 6 th edition breast-adjusted AJCC guidelines between 1998-2015 13 . Among the cases classified as Tis, we excluded those associated with lobular carcinoma in situ, Paget's disease, and an unknown or invasive histology type. We excluded cases with unknown laterality and unknown or no surgical intervention for the primary tumor. We also excluded patients who underwent mastectomy to study those who underwent lumpectomy only.
The flow chart of selection is provided in Figure 1.The research protocol was approved by the research ethics board of the Second Xiangya Hospital, Central South University, China. All methods were carried out in accordance with relevant guidelines and regulations. Since this is a retrospective study, the need for the informed consent was waived by The Second Xiangya Hospital. However, we submitted a data use agreement form to access the SEER Research Data File. Upon acceptance of the agreement, the SEER*Stat software and data files were downloaded directly from the SEER website.

Variables
The SEER data included patient age at diagnosis, race (White, Black, or Other (American Indian/AK Native or Asian/Pacific Islander)), year of diagnosis, laterality, pathologic grade, surgery of the primary site, radiation, estrogen receptor status, progesterone receptor status, tumor size and vital status, which are associated with patients' outcome. We determined 2 time intervals: the time from DCIS to death and the time between the diagnosis of BC and a second primary cancer incidence (contra-lateral breast cancer and ipsilateral breast cancer included). We divided all the patients diagnosed with a second primary BC after DCIS into ipsilateral and contralateral groups to compare the laterality of the two primary cancers. Patients eligible for study were categorized into 2 groups: lumpectomy without radiation and lumpectomy with radiation. The groups were compared for the variables above, and differences were evaluated using standardized differences.

Statistical Analysis
To analyze the association between OS and treatment modality, we performed a propensity score-based model. We calculated the propensity score using logit first an then patient were 1:1 matched for age at diagnosis, race (White, Black, or Other (American Indian/AK Native or Asian/Pacific Islander)), year of diagnosis, laterality, pathologic grade, estrogen receptor status, progesterone receptor status and tumor size based on their propensity score. Caliper matching was performed by matching participants who were within 0.2 times the standard deviation of their propensity score 14 . A standardized difference of greater than 0.1 was considered a meaningful imbalance between comparison groups 15 . P value was also shown after chi square test.
To determine the extent to which radiation was associated with decreased risk of death and MPC in these various subgroups, we compared the hazard ratios (HRs) in different treatment brought to the subgroups, and the results were shown in the forest plots .
The incidence of second primary cancer was compared with the expected incidence in the general population by calculating standardized incidence ratios (SIRs) and 95% confidence intervals (CIs). In this case, SIR in the general population is 1, and the incidence risk increases when SIR>1, decreases when SIR<1 respectively. SIRs were calculated overall and stratified by clinical parameters. The significance between SIR was analyzed using iri in Stata 14.
Cumulative survival curves after DCIS for each patient variable were constructed using the Kaplan-Meier method and were analyzed using the log-rank test in GraphPad Prism 7. To analyze both mortality and MPC and avoid the biases caused by competing events censored, we looked into competing risk model. Cumulative incidence function for competing risks data has been proposed and Fine and Gray test valued the difference using 'cmprsk' package. We also conducted Kaplan-Meier curves stratified by MPC and only one primary cancer group to observe the mortality difference. Age at diagnosis, race (White, Black, or Other (American Indian/AK Native or Asian/Pacific Islander)),pathologic grade, estrogen receptor status, progesterone receptor status and patients' prognosis were identified by Cox proportional hazards regression, and the associated hazard ratios (HRs) were reported. Statistical analysis was performed using R version 3.6.3 software (http://www.r-project.org/).

Baseline clinical characteristics
Among the 108,416 patients involved in the cohort, 39,039 patients were treated with lumpectomy alone, and 69,377 were treated with lumpectomy and radiotherapy ( Radiotherapy was more commonly used when the estrogen receptor status or the progesterone receptor status was positive. Abbreviations: SMD, standardized mean difference

Radiotherapy reduced the risk of mortality and MPC
In the matched comparison of patients treated with lumpectomy and radiotherapy vs,lumpectomy, radiotherapy was associated with a reduced risk of death (HR=0.7444, 95%CI: 0.7139-0.7761, p<0.0001) (Fig. 2a). In addition, we determined the time interval between the first diagnosis of DCIS and the occurrence of multiple primary cancer (MPC) among the 36,688 propensity-matched pairs. The adjusted HR for MPC in the lumpectomy and radiotherapy vs. lumpectomy alone groups was 0.801 (95%CI: 0.7728-0.8301) (Fig. 2b). Moreover, radiotherapy benefited disease-specific specific survival. (Fig. s1)As radiation reduces MPC but may increase MPC free mortality, competing risk analysis were used to look into. We compared breast cancer specific mortality competing death due to other causes (Fig.s2a) and MPC with competing risk of death without MPC (Fig.s2b). Lumpectomy plus radiotherapy reduced the cumulative incidence of breast cancer specific death and other tumor specific death. Meanwhile, radiotherapy benefited on reducing the risk of MPC, and MPC free mortality. Next, we assessed the protective effect of radiotherapy on mortality and MPC incidence in different subgroups based on the PSM model (Fig. 3). All subgroups could benefit from radiotherapy, but the degree of benefit varied among different subgroups. tumors benefited more than other subgroups in terms of mortality (Fig. 3a). ER negative patients could benefit more compared with those ER+ patients.

Effect of radiotherapy on the standardized incidence ratio differed by tumor type
Next, we calculated and compared the standardized incidence ratios (SIRs) among DCIS patients treated with lumpectomy vs. lumpectomy plus radiotherapy (Fig. 4).

Radiotherapy reduced the risk of occurrence of MPC rather than the mortality of MPC patients to reduce the overall mortality
To elucidate the association between radiotherapy and the prognosis of patients with only one primary cancer, KM curve analyses were performed. MPC patients had a higher risk of death than patients with only one primary cancer, and the HR was 1.944 (95%CI: 0.848-2.045) (Fig. 5a). Radiotherapy reduced the risk of death in only the group of patients with one primary cancer (Fig. 5c), and no statistically significant reduction in the risk of death was found in the MPC group between patients treated with lumpectomy alone vs. those treated with lumpectomy plus radiotherapy (Fig. 5b).

Discussion
Our study is the first to characterize the risk of MPC and death in DCIS patients in a large population using propensity score matching analysis. Among those with DCIS treated with lumpectomy, the majority of patients received radiotherapy, and these patients were middle-aged and elderly and were more likely to have high-grade disease, ER-positive status, PR-positive status and smaller tumor size than those who did not undergo radiation (Table 1). These factors were accounted for in the matched analysis, and propensity scoring was used to balance the baseline characteristics between the groups (Table s1). Besides, competing risk model minimalized the biases when competing events were censored. Therefore, we believe that the reductions in the risk of death and MPC are a result of the radiotherapy.
In the present study, the incidence of a second primary breast, kidney, renal pelvis and thyroid cancer as well as leukemia after DCIS increased, similar to previous studies 16,17 , possibly related to the mutated driven gene and endocrine therapy, specifically tamoxifen, had a slightly increased risk of uterine primary. As previously established, this increased risk of MPC is partially attributable to genetic and environmental influences. Importantly, we found that radiotherapy could reduce the risk of MPC by 20%, and several reports support our conclusion. Kaire Innos et al 18

439). A newest observational cohort study from the NHS Breast Screening
Programme and the National Cancer Registration and Analysis Service observed patients with breast-conserving surgery (BCS) + radiotherapy had lower risk of invasive breast cancer after DCIS compared to those performed BCS 21 . There are several possible underlying mechanisms: 1) There is statistical evidence of departure from a linear dose-response, with a downward curvature seen at high doses, likely resulting from radiation-induced cell sterilization 22 . 2) High-dose radiation may suppress estrogen production by relevant target cells, as well as androgen production by other cells (with androgen having the potential to be converted peripherally into estrogen) 23 . 3) Previously hidden tumor-associated antigens, stress proteins and danger-associated molecular patterns will be exposed after radiotherapy 24 . The exposure of antigens might activate antitumor immunity 25 and stimulate tumor cell killing.
In addition, we sought to identify those individuals who would benefit the most from radiotherapy. Lumpectomy combined with radiotherapy reduced the risk of ipsilateral second BC compared with lumpectomy alone. Among patients younger than 40 years, Black women with high-grade tumors were likely to benefit most. A possible reason could be that the second cancer risk is substantially greater among patients with initial diagnosis of breast cancer at the age of <50 years and high-grade tumors has a higher-grade malignancy and a greater accumulation of genetic mutations and epigenetic modifications which related to a second primary cancer and higher mortality. Therefore, these patients could benefit more due to their possible higher risk of mortality  27 . A recent population-based cohort study observed that the rate of ipsilateral invasive BC for women undergoing breast-conserving surgery with no record of radiotherapy was higher than that for women receiving radiotherapy (adjusted rate ratio 1.43, 95% confidence interval 1.05 to 1.96), and the mortality from BC between the various treatment groups was similar 21 .
In our study, we found that radiotherapy could reduce the risk of death, and the HR was 0.744 (95%CI: 0.7139-0.7761) among the cohort. However, the reduction in death only occurred in the patients with only one primary cancer, and radiotherapy did not have a statically significant effect among the MPC patients. We further compared mortality and found that the occurrence of MPC led to a higher risk of death than no occurrence of MPC. Therefore, radiotherapy reduced the risk of MPC rather than the death associated with MPC and thereby reduced overall mortality.

Limitations
Our study has several limitations. First, this is a retrospective study from the SEER database rather than a prospective cohort study, so inherent selection biases may undermine its external validity. Second, similar to other SEER database studies, this study lacks specific information on chemotherapy and hormone treatment. However, our PSM model controlled age, year of diagnosis, pathologic grade, hormone receptor status and tumor size, which determined the patients' systemic therapy usage. On the other hand, comparison between SIRs of second primary cancer after DCIS for patients treated with lumpectomy or lumpectomy plus radiotherapy showed that only statistical significance observed in second female breast cancer, demonstrated a relatively small impact on non-breast tumor brought by chemotherapy and hormone treatment. Third, the treatments for patients were not assigned at random. In addition, cases with some unknown variables were excluded; thus, the results may not include all patients with DCIS.

Conclusion
Among patients with DCIS, treatment with lumpectomy plus radiotherapy was associated with a significant reduction in mortality and risk of MPC, mainly second primary BC. Younger patients, Black women with high-grade tumors were likely to benefit most. The effect of the sequence of radiation and surgery and the form of radiotherapy should be explored in future studies.
Supplementary Materials: provided in the supplementary file.
Author Contributions: Yitong Li and Wenjun Yi finished the methodology and data curation, Yitong Li prepared the original draft preparation and Qitong Chen, Dengjie Ouyang and Taohong reviewed and edited the manuscript. Wenjun Yi visualized and supervised the research. All authors have read and agreed to the published version of the manuscript.
Funding: This research received no external funding.

Conflicts of Interest:
The authors declare no conflict of interest.