Clinicopathological features, treatment patterns, and prognosis of squamous cell carcinoma of the breast: an NCDB analysis

A total of 686 SCC patients and 1,211,403 IDC patients were identified from the database. Given the huge discrepancy of the amount of the SCC and IDC patients, we performed a 1:4 SCC-IDC matching on the following factors: Year of diagnosis (2004–2014), Facility type (Community Cancer Program, Comprehensive Community Cancer Program, Academic/Research Program, Integrated Network Cancer Program, Unknown), Facility location (New England, Middle Atlantic, South Atlantic, East North Central, East South Central, West North Central, West South Central, Mountain, Pacific), city type (Metropolitan, Non-metropolitan/Unknown), type of insurance (Not insured, Private insurance, Medicaid, Medicare, other Government, Unknown.). There were 686 SCC and 2744 IDC patients being selected as the final cohort for analysis. This study was an epidemiological study using de-identified data from the NCDB database. Therefore, consent for patient participation and study publication was not required. The study approval was waived by the ethical committee of Yale University and Sun Yat-sen Memorial Hospital.

The following data were collected for each patient: the year of diagnosis, age, race, Charlson-Deyo score, tumor grade, lymphovascular invasion, T-stage, N-stage, histology, estrogen receptor (ER) status, progesterone receptor (PR) status, HER2 status, primary surgery categorization, radiation therapy (RT), chemotherapy, neoadjuvant chemotherapy, response to chemotherapy, endocrine therapy, survival month and OS status. Patients were categorized into three age groups based on their ages at diagnosis (≤50 yrs., 50–60 yrs., > 60 yrs). We used this cut-off because the median age was close to 60 yrs. in our study population and because 50 yrs. is the usual cut-off age for premenopausal and post-menopausal women. Histology was divided into two categories, namely, IDC and SCC.

We conducted a descriptive analysis of the baseline clinicopathological features of the included patients and used the Chi-square test to compare the characteristics of the patients with different histologies. The median follow-up time was calculated as the median observed survival time of the entire population. OS was measured as the time from diagnosis to death due to any cause. The cumulative OS rates were estimated using Kaplan-Meier analysis. We used a Cox regression model to screen for prognostic factors of OS. We tested the proportional hazards assumption by plotting the scaled Schoenfeld residuals of all coefficients over time and found no violations. All P-values were two-sided. P-values less than 0.05 were considered statistically significant. Statistical analysis was performed using Stata/MP, version 13.0 (StataCorp LP, College Station, TX, USA).

In patients with T1–2 stages who did not received neoadjuvant chemotherapy, there were 58.3% (120/206) vs. 65.4% (890/1362) of the SCC and IDC patients received BCS (P = 0.048) respectively. Among the patients with BCS (N = 1791, 16 patients with unknown RT status were excluded.), 65.3 and 83.0% of the SCC and IDC patients received RT (P < 0.01), respectively. The use of RT in node-positive patients with mastectomies, were similar in SCC and IDC patients (42.1% (37/88) vs. 48.4%(166/343), P = 0.287).

In patients with favorable prognosis (hormone receptor (HR)-positive, HER2-negative and node-negative), chemotherapy was performed in 42.9% (12/28) and 18.7% (80/427) of the SCC and IDC patients, respectively (P = 0.002). In the HR-positive, node-negative patients who were diagnosed before 2010 (when the HER2 status was unknown), there were 51.6% (32/62) and 24.9% (248/997) of the SCC and IDC patients who had received chemotherapy, respectively (P < 0.001). Among the patients with HR-positive disease, endocrine therapy was performed in 51.6% (79/153) and 70.5% (1446/2050) of the SCC and IDC patients, respectively (P < 0.001).

In this study, there were 298 patients with known history of neoadjuvant chemotherapy, and 129 of them had clear information about treatment responses (CR, PR, CR/PR, No response). SCC (vs. IDC) was significantly associated with no responses to neoadjuvant chemotherapy (20% vs. 5.05%, P = 0.019) (Table 2).

With a median follow-up time of 58.3 months, the respective 5-yr and 10-yr OS were 62.1 and 50.6% for the SCC patients, and 83.0 and 69.5% for the IDC patients, respectively (P < 0.001). SCC (vs. IDC) was associated with poorer OS in univariate analysis (HR = 2.39, 95%CI 2.06–2.77, P < 0.001), and in multivariate (HR = 1.40,95%CI 1.17–1.67, P < 0.001) analysis after adjusting for age, race, comorbidity, T-stage, N-stage, M-stage, ER, PR, tumor grade, LVI, surgery, endocrine therapy, chemotherapy and RT (Table 3, & Additional file 1: Figure S1). In patients who were diagnosed after 2010, SCC (vs. IDC) was still associated with poorer OS (HR = 1.57, 95%CI 1.11–2.21, P = 0.011), after adjusting for the above variables, as well as LVI and HER2 status.

We hypothesized that there was interaction between hormonal status (positive vs. negative) and the histology (SCC vs. IDC) in the analysis of OS. We performed a subgroup analysis and noticed that positive (vs. negative) hormonal status was significantly associated with improved OS in IDC (P < 0.01) patients, but not in the SCC (P = 0.042) patients (Fig. 1, Interaction test, P = 0.023). However, endocrine therapy was also associated with improved OS for the HR-positive IDC patients (HR = 0.61, 95%CI 0.48–0.77, P < 0.001), as well as for the HR-positive SCC patients (HR = 0.30, 95%CI 0.15–0.59, P < 0.001) (Fig. 2), after adjusting for age, race, comorbidity score, grade, T-stage, N-stage, M-stage, and chemotherapy.

Previous studies have shown that SCC of the breast is more likely to be HR-negative and is associated with worse clinical outcomes [2, 8, 9]. Hennessy et al. [2] reported that the 5-year overall survival (OS) rates of 33 and 137 SCC patients selected from the M.D. Anderson Cancer Center and the SEER database, respectively, were 40 and 64%, respectively. In an update, Yadav et al [9] reported a 5-year cancer-specific survival rate of 63.5% for 445 SCC patients who were collected from the SEER database. Due to the rarity of SCC, only the data from the national cancer database is able to provide adequate statistical power to study the prognosis of SCC. However, several important prognostic factors, and the information of adjuvant therapies were lacking in the SEER database. In contrast, the NCDB database provides more prognostic factors (e.g comorbidity score, HER2 status and LVI status), as well as the information of adjuvant therapies (chemotherapy and endocrine therapy), therefore allows a more accurate estimation with less bias. In this study, we observed that the SCC patients had poorer clinicopathological features (e.g., T3–4, grade III, HR-negative disease) than the IDC patients. The adjusted analysis demonstrated that the SCC patients had significantly worse clinical outcomes than the IDC patients in both the 2004–2014 and 2010–2014 (HER2 status adjusted) cohorts. In consistent with previous studies, we confirmed that SCC (vs. IDC) is associated with poorer OS, after adjusting for more prognostic factors and adjuvant therapies.

The concept that SCC has worse clinical outcomes than IDC is likely to affect the choice of treatments in the clinical practices. This is confirmed in our study that chemotherapy was more prevalent in low-risk (HR+/HER2-/LN-) SCC (vs. IDC) patients (42.9% vs. 18.7%, P = 0.002). However, whether the SCC is responsive to chemotherapy is still unknown. Hennessy et al. [2] and Zhang et al. [6] reported that they used no responses were observed in their SCC patients after neoadjuvant chemotherapy. Only two case-reports [10, 11] have suggested that cisplatin-based chemotherapy is able to achieve long-term control, but these results need further verification. In our study, we noticed that SCC (vs. IDC) was significantly associated with no responses to neoadjuvant chemotherapy (20% vs. 5.05%, P = 0.019). Therefore, the benefit of chemotherapy in SCC patients remains unknown.

In this study, we observed that the endocrine therapy was less performed in the HR-positive SCC (51.6%) patients, than in the HR-positive IDC (70.5%) patients. A contributing reason could be that endocrine therapy in the head & neck or esophageal SCC patients is not useful in clinical practices [12, 13], even if the in vitro evidences [14, 15] had suggested the role of tamoxifen for SCC of the oral cavity or esophagus. However, SCC of the breast might possibly be different from the head & neck SCC. A recent study used whole-exome sequencing to show that the SCC components have nearly identical landscapes of somatic mutations to their adjacent IDC component, suggesting that SCC may originate from the IDC [3]. Since the role of endocrine therapy had been established in IDC patients, it is possible the endocrine therapy would also be beneficial in SCC patients. In the analysis of prognostic factors of OS, we observed a significant interaction between the HR status and the histology (SCC vs. IDC). The HR status was prognostic only for IDC patients, but not for SCC patients. Despite of this, we still observed that the endocrine therapy significantly improves the OS in HR-positive IDC and SCC patients. Taken together, we suggested that endocrine therapy should remain as the standard treatment for HR-positive SCC patients. Ng et al. [5] studied the landscape of somatic genetic alterations of SCC and reported that TP53(78%) and PI3KCA(44%) are the most frequently mutated genes in SCC. They proposed that the mutation affecting genes might result in the Wnt and mTOR pathway activation. Future studies are warranted to investigate whether relevant pathway inhibitors could be used for SCC patients.

We assessed the influence of the SCC component on the local therapy. SCC patients had a slightly lower rate of BCS, when compared to IDC patients (58.3% vs. 65.4%, P = 0.048). The underlying reason could be that the SCC patients had larger tumor than IDC patients (T3–4: 23.1% vs. 4%). Among patients with BCS, RT was used in 65.3 and 83.0% of the SCC and IDC patients, respectively. Currently, evidence that support the use of RT for SCC is lacking. Hennessy et al. [2] reported that 4 out of 19 SCC patients treated with RT had locoregional relapse within the irradiated field, suggesting that SCC might be radioresistant. Two studies [16, 17] reported no benefit of RT on OS, but the small sample sizes in these studies limited their statistical power. Using the SEER database, Wu et al. [18] reported that RT was significantly associated with improved OS but not cancer-specific survival, which is difficult to explain. Furthermore, they reported that RT was significantly associated with improved CSS in stage II SCC patients, but the analysis was not adjusted for ER, PR or HER2. Thus, the role of RT as an adjuvant local control therapy after surgery remains controversial.

First, it is possible that the IDC patients may have a small proportion of SCC component area. Without pathological confirmation, grouping the cases into IDC, and SCC might not be always accurate. However, a detailed histopathology examination is impossible in mining large database, such as NCDB. The large sample size of this study population is able to compensate this limitation. Second, nonrandomized comparisons of treatment effects are prone to providing misleading estimations. One study [19] showed that the treatment effect of RT in breast cancer was over-estimated in observational data compared with randomized clinical trial data. This effect is reasonable as the “treatment-by-indication” bias can never be eliminated in observational data. Therefore, we did not estimate the survival benefit of chemotherapy, as patients with more advanced diseases are prone to receive chemotherapy. However, estimation of the survival benefit of endocrine therapy in HR-positive patients is less likely to be affected by the “treatment-by-indication” bias. HR-positive is the only indication for endocrine therapy. In addition, the decision to implement the endocrine therapy is less likely to be influenced by the comorbidity status. Thus, estimation of the survival benefit of endocrine therapy using observational data is reasonable. Due to the rarity of SCC of the breast, a prospective, randomized study for SCC cannot be realistically implemented. Third, the NCDB did not have information regarding trastuzumab therapy. Whether trastuzumab would be appropriate for HER2-positive SCC patients remains unclear. The lack of information regarding local relapse, metastatic relapse and cancer-specific survival is also one of the limitations. Fourth, there is a growing awareness of an association between SCC of the breast and implants. But the significance of the association is unclear due to the rarity of this situation [20‐22]. The NCDB database does not have the information about the history of breast implants augmentation before the diagnosis of SCC, therefore limits our understanding about this issue.