Risk of breast cancer-related death in women with a prior cancer.

The overall risk of developing a second primary cancer is increasing. The purpose of this study was to analyze the survival of patients with breast cancer diagnosed after a prior cancer and identify risk factors of breast cancer death in this population. Using the SEER database, we identified 1,310 woman diagnosed with breast cancer between 2010 and 2015 after a prior cancer as the primary cohort. Clinicopathological characteristics were compared using the Student t-test and chi-square test. Fine and Gray’s regression was used to evaluate the effect of treatments on breast cancer death. After propensity score matching (PSM), 9,845 pairs of patients with breast cancer as the prior or second cancer diagnosed between 2010 and 2011 were included as a second cohort. PSM-adjusted Kaplan-Meier and Cox hazards models were used to evaluate the impact of prior cancer on survival. The results showed that survivors of gynecologic cancers (e.g., ovarian cancer) had a higher risk of developing breast cancer than survivors of gastrointestinal and urinary tract cancers. More patients died of breast cancer than of prior urinary cancer (53.3% vs. 40%, P < 0.05) and melanoma (66.7% vs. 33.3%, P < 0.05). The ratio of breast cancer deaths to prior cancer deaths was significantly higher in patients with diagnoses interval ≥ 3 years than in those with the interval < 3 years (2.67 vs. 0.69, P < 0.001). Breast cancer-specific survival and overall survival rates were significantly lower in women with breast cancer as the second primary cancer than in those with breast cancer as the prior cancer, especially among hormone receptor-positive women. However, breast cancer treatment decreased the risk of breast cancer -specific death (hazard ratio = 0.695, 95% confidence interval: 0.586–0.725, P < 0.001). Breast cancer patients with prior cancers must be carefully considered for clinical trials.


INTRODUCTION
Due to the advances in the detection of early-stage cancers and cancer treatment, the population of cancer survivors has increased by approximately 4 folds in the United States in the past 30 years [1][2][3]. Almost twothirds of cancer survivors live more than 5 years after the initial diagnosis, which increases the risk of developing a second primary malignancy (SPM) [4][5][6][7][8]. A SPM is defined as a cancer which arises independently in a new organ or tissue at least 2 months after the diagnosis of the prior primary cancer. Over 10% of younger adult cancer patients and around 25% of older adult cancer patients have been diagnosed with a SPM [9,10]. It has been reported to be associated with genetic susceptibility [11,12], the carcinogenic AGING adverse effects of cancer treatment [13], and behavioral risk factors such as smoking and alcohol intake [14][15][16][17][18].
Many studies have examined the risk of a SPM in patients with different types of prior cancer, such as lung, testis, head and neck, and thyroid cancers [19][20][21]. The prognosis of patients with stage IV lung cancer as a SPM was not affected by a prior cancer [22]. The prevalence of colorectal adenomas in breast cancer survivors is similar to that in patients with single colorectal adenomas [23]. Garg et al. [24] reported that early diagnosis and the absence of recurrence in patients with prior breast cancer who had abdominal carcinomatosis were significantly associated with the development of ovarian/peritoneal cancer as a SPM. Prior breast cancer and tamoxifen exposure did not affect the prognosis of women with uterine papillary serous carcinoma as a SPM [25]. However, the risk of breast cancer as a SPM in patients with a prior cancer and the cancer-specific survival for these patients are not known.
Thus, in the present study, we used data from the Surveillance, Epidemiology, and End Results (SEER) database to analyze the outcomes of patients who developed breast cancer as a SPM. This information may help guide long-term surveillance strategies for patients.

Baseline characteristics of the primary cohort
The median (interquartile range, IQR) age at diagnosis of the prior cancer was 66    (Table 1).

Breast cancer-related deaths in the primary cohort
OS varied significantly among patients with different types of prior cancer (P < 0.001), and patients with prior lung cancer had the shortest OS ( Figure 1A). We stratified patients by death of prior cancer and death of breast cancer as an SPM. Overall, 30.5% of patients died of breast cancer, and 40.2% of patients died of the prior cancer. The breast cancer-related death rate was the lowest (19.5%) in patients with prior gastrointestinal cancer and the highest (66.7%) in patients with prior melanoma. Breast cancer caused more deaths than prior urinary cancer (53.3% vs. 40%) and melanoma (66.7% vs. 33.3%), but caused less deaths than prior lung cancer (26.8% vs. 43.9%), hematological cancer (35.3% vs. 64.7%), and gynecologic cancer (26.8% vs. 36.6%) (all P < 0.05) ( Figure 1B). The ratio of breast cancer deaths to prior cancer deaths was markedly higher in patients with diagnoses interval ≥ 3 years than in those with the interval < 3 years (2.67 vs. 0.69, P < 0.001).

Associations of demographic and clinicopathologic factors with breast cancer deaths in the primary cohort
The rates of high-grade disease (32.3% vs. 23.1%, P < 0.001), T3-4/N0/M0 disease (11.3% vs. 4.4%, P = 0.014), and distant disease (33.9% vs. 27.5%, P = 0.023) of breast cancer were significantly higher in patients who died of breast cancer than in those who died of a prior cancer, and the rates of metastasis of prior cancer (11.3% vs. 25.3%, P = 0.011) and administration of breast cancer treatment (37.1% vs. 63.4%, P = 0.018) were significantly lower in patients who died of breast cancer. Patients who died of breast cancer had a longer diagnosis interval (17.0 vs. 11.4 months, P = 0.040) and were older (73.5 vs. 69.1 years, P = 0.035) than those who died of a prior cancer ( Table 2). Breast cancer treatment was associated with a decreased risk of breast cancer-specific mortality (BCSM) (hazard ratio = 0.695, 95% confidence interval [CI] = 0.586-0.725, P < 0.001), but not associated with a decreased risk of non-BCSM ( Figure 2).

Risk of breast cancer-related deaths among patients with different types of prior cancer in the primary cohort
To further determine the risk of breast cancer-related deaths among patients with different types of prior cancer, we calculated the ratio of breast cancer deaths to prior cancer deaths in patients with low-grade and low stage(stage cT1-2/N0/M0) and in those with high-grade or high stage(stage cT3-T4/N0/M0) cancer ( Figure 3). Overall, the ratio was 0.47 in patients with low-grade and stage cT1-2/N0/M0 disease and 1.31 in patients with high-grade or stage cT3-T4/N0/M0 breast cancer, indicating that the risk of breast cancer deaths was much higher than that of prior cancer deaths in the latter group of patients. In total, 51.8% of patients with high-grade or stage cT3-4/N0/M0 breast cancer were likely to die of breast AGING Interestingly, patients with prior oral cancer were more likely to die of breast cancer, whereas those with prior lung, hematological, and gynecological cancers were more likely to die of the prior cancer. However, taking tumor grade and stage into consideration, patients with prior gastrointestinal cancer, melanoma, and urinary tract cancer were more likely to die of breast cancer when they had high-grade or stage cT3-T4/N0/M0 breast cancer, but were more likely to die of prior cancer when they had low-grade and cT1-2/N0/M0 breast cancer.

Survival of patients with breast cancer as the prior cancer or subsequent primary cancer in the second cohort
In the second cohort, 63,761 patients had breast cancer as their only cancer (primary breast cancer, PBC), and 9,955 had breast cancer as the second primary cancer (subsequent breast cancer, SBC). The 5-year BCSS and OS rates were significantly lower in patients with SBC than in those with PBC (BCSS: 91.0% vs. 94.0%, P < 0.001, Figure 4A; OS: 91.0% vs. 93.8%, P < 0.001, Figure 4B).   Table 3). The survival of the matched groups was in consistent with the survival of the entire cohort, the 5-year BCSS and OS rates were significantly lower in patients with SBC than in those with PBC (BCSS: 91.2% vs. 93.6%, P < 0.001, Figure 4C; OS: 80.5% vs. 86.1%, P < 0.001, Figure 4D).

Survival of patients with hormone receptor (HR)positive breast cancer as the prior cancer or second primary cancer
Taking hormone receptor statuses into consideration, breast cancer as the subsequent primary cancer (SBC) was significantly associated with short BCSS and OS in HER2-/HR+ patients (both P < 0.001, Figure 5A, 5B) and HER2+/HR+ patients (P = 0.001and P < 0.001) ( Figure 5E, 5F). However, no such associations were observed in HER2+/HR-and triple-negative patients ( Figure 5C, 5D, 5G, 5H).

Factors associated with survival of patients with breast cancer as the primary cancer or subsequent breast cancer
We used univariate and multivariate analyses to determine clinicopathological factors which associated with survival. Univariate analysis showed that age, race, marital status, tumor differentiation, TNM stage, breast cancer subtype, breast cancer as the prior or second  BCSS was significantly shorter in patients with breast cancer as the second primary cancer than in those with breast cancer as the prior cancer in the entire cohort. (B) OS was significantly shorter in patients with breast cancer as the second primary cancer than in those with breast cancer as the prior cancer in the entire cohort. After PSM, both BCSS (C) and OS (D) were significantly lower in patients with breast cancer as the second primary cancer than in those with breast cancer as the prior cancer. , and radiotherapy were independent prognostic factors for both BCSS and OS, race was an independent prognostic factor for BCSS, and age and chemotherapy were independent prognostic factors for OS (all P < 0.05) ( Table 4).

AGING
As shown on Figure 6, compared with breast cancer as the primary cancer (PBC), breast cancer as the SBC was associated with shorter BCSS when the age at diagnosis was > 65 years (hazard ratio =

DISCUSSION
The cancer survivor population is rapidly growing, and subsequently the number of patients with multiple primary or multi-organ cancers is also increasing [26]. In the US, the cancer survivor population presented a 2% annual increase, and approximately 18% of cancer cases were developed after the prior cancer according to SEER data [27]. The risk of developing an SPM among cancer survivors was reported to be as high as that in the general population [28]. The risk of breast cancer as a second primary cancer is especially high [29]. However, the clinicopathological characteristics and survival of patients with this disease remain largely unknown.
In this population-based study, we found that gynecologic cancers, gastrointestinal cancer, urinary tract cancers, hematological cancers, lung cancer, and melanoma were the most common types of prior cancer. Our results showed that breast cancer as the second primary cancer was associated with short survival. The 5-year BCSS and OS rates were significantly lower in patients with breast cancer as the second primary cancer than in those with breast cancer as the prior cancer in both the standard and PSM analyses, these results are   [32]. A subgroup survival analysis also showed that HER2−/HR+ and HER2+/HR+ patients with breast cancer as the second primary cancer had shorter BCSS and OS than other subgroup. As such, breast cancer patients with prior cancers must be carefully considered for clinical trials. Univariate and multivariate analyses indicated that breast cancer as the second primary cancer was an independent risk factor for BCSS and OS. The high occurrence and death rates of breast cancer in cancer survivors may likely be attributed to BRCA1 and BRCA2 mutations as well as chemotherapy and radiotherapy for prior cancer. Patients who received underwent chemotherapy for prior cancer may not respond well to systemic treatment for their second primary cancer due to decreased efficacy and tolerability after prior systemic treatment [33,34]. Interestingly, Fine and Gray's regression analysis showed that breast cancer treatment was associated with a decreased risk of BCSM. The ratio of breast cancer deaths to prior cancer deaths was significantly higher in patients with diagnoses interval ≥ 3 years than in those with the interval < 3 years. Although patients' breast cancer as the second primary cancer had short survival, breast cancer treatment can still prolong survival.
Some limitations in the present study need to be mentioned. First, as a retrospective study, selection bias was inevitable. Although we performed PSM to AGING minimize such bias, a PSM analysis is also vulnerable to hidden biases, and residual confounding factors could not be entirely ruled out. Second, as the data were retrieved from the SEER database, we could not control treatment-related factors of the prior cancer, which may alter findings related to second cancer incidence and survival. Therefore, further studies with more data are required to validate our findings.

CONCLUSIONS
Survivors of gynecologic cancers such as ovarian cancer tend to have a high risk of breast cancer. Breast cancer as a second primary cancer, especially locally advanced or high-grade breast cancer, is a significant cause of death in patients with a prior cancer. However, breast cancer treatment decreased the risk of BCSM in these patients. Considering the increases in cancer survivors and deaths related to breast cancer as the second primary cancer, effective detection and treatment strategies is warranted to be investigated in this population.

MATERIALS AND METHODS
The SEER database is a population-based cancer registry sponsored by the US National Cancer Institute (NCI). The SEER program collects data regarding patient demographic characteristics, cancer incidence, treatment, and survival. The current study included 2 independent patient cohorts derived from the SEER database for separate analyses.

Primary cohort
The principal analysis of this study used the multiple primary-standard incidence ratio function of SEER*Stat version 8.  [35,36].
The propensity score matching (PSM) method with a ratio of 1:1 was performed to balance the baseline characteristics of patients with breast cancer as the second primary cancer and as the prior cancer [37]. Breast cancer-specific survival (BCSS) was defined as the duration from the diagnosis of breast cancer to death due to breast cancer. Overall survival (OS) was defined as the duration from the diagnosis of breast cancer to death from any cause. Survival between groups was AGING compared using the Kaplan-Meier method. Logistic regression analyses were used to examine the relations between clinicopathological characteristics. A 2-sided P value of < 0.05 was considered significant.

AUTHOR CONTRIBUTIONS
F.J. and C.Q.Y. conceived and designed the study; M.Y. and L.L.Z. analyzed the data; F.J. drafted the paper; T.Z., M.Y.C., Q.Q.X., S.Y.W., A.L.Z., K.W. reviewed the manuscript. All authors interpreted the data, edited or commented, and approved the final manuscript.

CONFLICTS OF INTEREST
All the authors declare that they have no conflicts of interest.