Clinical characteristics and prognosis of primary small cell carcinoma of the breast: a propensity score-matched, population-based study Authors: Lei Yin, Ai-Hong Yin, Chen-Chen Pu Published: 2023 ABSTRACT AbstractObjectiveThe purpose of this study was to describe the clinicopathological characteristics and prognosis of primary small cell carcinoma of the breast (PSCCB) and compare PSCCB with breast invasive ductal carcinoma (IDC).DesignA retrospective cohort study.SettingData of patients with PSCCB and breast IDC were identified from the Surveillance, Epidemiology, and End Results (SEER) database between 2004 and 2016.ParticipantsEighty-three patients with PSCCB and 410 699 patients with breast IDC were enrolled in the present cohort study.Materials and methodsPatients with PSCCB and breast IDC were identified from the SEER database between 2004 and 2016. The clinicopathological characteristics and survival of patients with PSCCB and IDC were compared. Propensity score matching (PSM) analysis was performed to adjust for differences in baseline characteristics when comparing overall survival (OS) and cancer-specific survival (CSS). Moreover, OS-/CSS-specific nomograms were established to predict the prognosis of PSCCB.ResultsCompared with IDC, PSCCB was significantly correlated with older age, male, higher pathological grade, higher TNM (tumour, node, metastases) stage, a higher proportion of triple-negative breast cancer, a lower proportion of ER/PR positivity and significantly worse clinical outcome. The median OS and CSS of patients with PSCCB were 23.0 m (95%CI 13.0 to 56.0) and 28.0 m (95%CI 18.0 to 66.0), respectively. The 5-year OS and CSS rates in the PSCCB group were 36.1% and 42.4%, respectively. In the matched cohort after PSM analysis, patients with PSCCB had significantly worse OS and CSS than IDC patients. Multivariate Cox regression analysis demonstrated that T stage and administration of chemotherapy were independent prognostic factors for both OS and CSS in patients with PSCCB. The C-index for OS-/CSS-specific nomogram was 0.75 (95%CI 0.66 to 0.85)/0.79 (95%CI 0.69 to 0.89), respectively. The calibration curve in the ROC analysis indicated that the predicted value was consistent with the actual observation value. Decision curve analysis suggested that the nomogram model has a significant positive net benefit from the risk of death and are better than the traditional TNM staging system.ConclusionPSCCB has distinct clinicopathological characteristics, and patients with PSCCB have significantly worse clinical outcomes than those with IDC.Keywords: Breast tumours, Prognosis, ONCOLOGY Strengths and limitations of this study This is the largest data set of patients with prognosis of primary small cell carcinoma of the breast (PSCCB) to describe its demographic, clinicopathological features and prognosis using a population-based approach. This is the largest data set of patients with prognosis of primary small cell carcinoma of the breast (PSCCB) to describe its demographic, clinicopathological features and prognosis using a population-based approach. This study is the first investigation to compare the difference in clinicopathological characteristics and survival between PSCCB and breast invasive ductal carcinoma (IDC). This study is the first investigation to compare the difference in clinicopathological characteristics and survival between PSCCB and breast invasive ductal carcinoma (IDC). To reduce the influence of uneven baseline characteristics on survival differences between PSCCB and IDC patients, this study compared the survival in the matched cohort using propensity score matching analysis. To reduce the influence of uneven baseline characteristics on survival differences between PSCCB and IDC patients, this study compared the survival in the matched cohort using propensity score matching analysis. This Surveillance, Epidemiology, and End Results database-based retrospective study design limited the collection and analysis of several variables that may be relevant to survival. This Surveillance, Epidemiology, and End Results database-based retrospective study design limited the collection and analysis of several variables that may be relevant to survival. Introduction Small cell carcinoma (SCC) is a poorly differentiated neuroendocrine carcinoma with a low incidence, high mortality and high distant metastasis rate.1 SCC predominantly originates in the lung, but 5% of SCCs are found in extrapulmonary sites, such as the gastrointestinal tract, breast and cervix.2 Extrapulmonary SCC (EPSCC) is histologically similar to small cell lung cancer, including morphological and immunohistochemical features, and has rapid local progression and distant metastasis.3 Although primary SCC of the breast (PSCCB) accounts for approximately 4%–10% of all extrapulmonary SCCs, it is a very rare and aggressive subtype of breast cancer that accounts for less than 0.1% of breast carcinomas.4 5 Because of its rarity, all previous studies on PSCCB were case reports and small case series.6–8 These existing evidence suggest that PSCCB typically presents as a palpable breast or axillary mass in women over the age of 60, with 50%–67% of patients presenting with lymph node metastases.9 10 Previous studies also reported that hormone receptor positivity varied among PSCCB patients, ranging from 33% to 50%.7 11 However, this description of PSCCB comes from a small-sample cohort and requires validation in a large cohort. In addition, the standard therapeutic regimen for PSCCB is largely uncertain because of the small number of reported cases. In the previous literature, most patients were managed similarly to patients with breast invasive ductal carcinoma (IDC) based on the tumour stage, including surgery, radiotherapy and chemotherapy. However, available data suggest that patients with PSCCB often have poorer prognosis than patients with IDC, depending on the initial stage at diagnosis.12 13 Further, because follow-up studies with information on prognosis are limited, factors affecting survival in these patients remain unclear. Additionally, PSCCB may have different biological characteristics than breast IDC because of its specific histological features. Therefore, the similarity and difference in demographic features, clinicopathological characteristics, and prognosis between PSCCB and breast IDC are not well defined. To provide accurate information on the characteristics of PSCCB and breast IDC, we conducted a retrospective, real-world study on all PSCCB patients registered in the Surveillance, Epidemiology, and End Results (SEER) database from 2004 to 2016. This study aims to (1) describe the clinicopathological characteristics of PSCCB, (2) conduct a comparative analysis on the clinicopathological characteristics of PSCCB and IDC, (3) compare the prognosis of PSCCB and IDC using the propensity score-matched analysis, (4) identify potential prognostic clinicopathological factors of PSCCB and (5) further establish the prognostic nomogram. Materials and methods Patient and public involvement Data on patients suffering from breast cancer were extracted from the SEER database, as previously de-identified by the SEER. Patients This retrospective study used data from the SEER database, which consists of nine population-based cancer registries. The SEER Programme of the National Cancer Institute (NCI) routinely compiles and publishes national data on cancer patient demographics, primary tumour site, clinicopathological features and survival statistics, covering an estimated 34.6% of the US population.14 SEER is a publicly available database; therefore, neither institutional review board approval nor informed consent was required for this study. The SEER database was searched to identify all patients with PSCCB and IDC from 2004 to 2016 based on the International Classification of Diseases for Oncology, third ed. (PSCCB: ICD-0–3 8041/8042/8043/8044/8045; IDC: ICD-0–3 8500). The clinicopathological characteristics and survival data of each patient were isolated using SEER*Stat 8.3.9.2. Patients were included in the present study if the following criteria were met: PSCCB or IDC as the only primary cancer diagnosis and known information on survival time and survival status. PSCCB and IDC patients lacking primary tumour and survival data were excluded. The patient inclusion flowchart is shown in online supplemental figure 1. bmjopen-2023-073841supp001.pdf Clinicopathological characteristics To describe the clinicopathological characteristics of PSCCB patients, the following data fields were retrieved: age, sex, race, laterality, pathological grade, American Joint Committee on Cancer tumor, node, metastasis (AJCC-TNM) stage, T stage, lymph node and distant metastases, molecular subtype, estrogen receptor (ER) status, progestern receptor (PR) status, surgery, radiotherapy, chemotherapy, survival months, vital status and cause of death classification. To further compare these characteristics between PSCCB and IDC, some variables were coded and regrouped as follows: age was categorised as <60 years and ≥60 years, and the surgical method was divided into mastectomy or breast-conserving surgery. Overall survival (OS) and cancer-specific survival (CSS) were defined as the study endpoints in the present study. OS was defined as the interval from the initial diagnosis to death owing to any cause or the last follow-up, and CSS was defined as the interval from the initial diagnosis to death based on PSCCB/IDC or the last follow-up. Statistical analysis The clinicopathological characteristics are presented as numbers with percentages and compared between PSCCB and IDC using the χ2 test. Survival analysis was conducted using a Kaplan–Meier plot and log-rank test to determine significant differences in OS and CSS. Univariate and multivariate Cox regression analyses were used to identify prognostic factors of PSCCB and their weights, and only significant factors from the univariate Cox analysis were included in the multivariate Cox analysis. Moreover, OS-/CSS-specific nomograms were established to predict the prognosis of PSCCB. Area under the curve (AUC), C-index and calibration curve were used to determine the predictive accuracy and discriminability of nomograms. Decision curve analysis (DCA) was used to further compare the clinical effectiveness of the nomogram and TNM (tumour, node, metastases) staging. Because uneven baseline characteristics may greatly impact OS and CSS differences between PSCCB and IDC patients, a matched cohort was obtained using propensity score matching (PSM) analysis to eliminate baseline variations between these two groups. PSM analysis (1:5) was conducted to calibrate the effects of baseline clinicopathological characteristic differences between patients with PSCCB and IDC, including age, sex, race, laterality, grade, stage, T stage, N stage, M stage, molecular subtype, ER/PR status, surgery, radiation and chemotherapy. Survival analysis between PSCCB and IDC patients was conducted in the unmatched cohort (n=83 for PSCCB and n=4 10 652 for IDC) and matched cohort (n=83 for PSCCB and n=415 for IDC). All survival analyses were performed using R statistical software (v.4.1. http://www.R-project.org/) with packages ‘stringr’, ‘MatchIt’, ‘tableone’, ‘surviminer’, ‘ggDCA’, ‘ggsci’, and ‘survival’. Statistical significance was defined as p<0.05. Results Comparison of clinicopathologic characteristics between PSCCB and IDC patients Eighty-three patients with PSCCB and 410 699 with breast IDC were identified in the SEER database between 2004 and 2016 (online supplemental figure 1). The clinicopathological characteristics of the included patients are presented in table 1. There were significant differences between patients with PSCCB and IDC, including age, sex, grade, AJCC-TNM stage, T stage, M stage, molecular subtype, ER/PR status, surgery and chemotherapy. Table 1Baseline characteristics of patients with PSCCB and IDCCharacteristicsPSCCB (n=83)IDC (n=410 699)p valueAge <60 years33 (39.8%)211 595 (51.5%)0.04 ≥60 years50 (60.2%)199 104 (48.5%)Sex Female79 (95.2%)407 519 (99.2%)<0.01 Male4 (4.8%)3180 (0.8%)Race White55 (66.3%)321 150 (78.2%)<0.01 Black20 (24.1%)469,16 (11.4%) Other6 (7.2%)39 271 (9.6%) Unknown2 (2.4%)3362 (0.8%)Laterality Right45 (54.2%)201 743 (49.1%)0.30 Left37 (44.6%)208 425 (50.7%) Unknown1 (1.2%)531 (0.1%)Grade I0 (0%)75 917 (18.5%)<0.01 II5 (6.0%)161 904 (39.4%) III50 (60.2%)153 036 (37.3%) IV12 (14.5%)2608 (0.6%) Unknown16 (19.3%)17 234 (4.2%)Stage I8 (9.6%)175 883 (42.8%)<0.01 II28 (33.7%)124 244 (30.3%) III17 (20.5%)41 923 (10.2%) IV17 (20.5%)17 980 (4.4%) Unknown13 (15.7%)50 669 (12.3%)T stage T00 (0%)305 (0.1%<0.01 T113 (15.7%)215 807 (52.5%) T231 (37.3%)110 088 (26.8%) T317 (20.5%)18 577 (4.5%) T49 (10.8%)15 986 (3.9%) Unknown13 (15.7%)49 936 (12.2%)N stage N037 (44.6%)239 124 (58.2%)<0.01 N123 (27.7%)91 099 (22.2%) N25 (6.0%)22 254 (5.4%) N36 (7.2%)12 430 (3.0%) Unknown12 (14.5%)45 792 (11.1%)M stage M058 (69.9%)351 309 (85.5%)<0.01 M117 (20.5%)17 980 (4.4%) Unknown8 (9.6%)41 410 (10.1%)Molecular subtype HR+/HER2- (Luminal A)14 (16.9%)156 871 (38.2%)<0.01 HR+/HER2+ (Luminal B)0 (0%)28 941 (7.0%) HR-/HER2+ (HER2 enriched)0 (0%)12 868 (3.1%) HR-/HER2- (Triple Negative)21 (25.3%)29 070 (7.1%) Unknown48 (57.8%)182 949 (44.5%)ER status Negative53 (63.9%)85 149 (20.7%)<0.01 Positive12 (14.5%)310 388 (75.6%) Borderline0 (0%)470 (0.1%) Unknown18 (21.7%)14 692 (3.6%)PR status Negative51 (61.4%)125 790 (30.6%)<0.01 Positive12 (14.5%)267 007 (65.0%) Borderline0 (0%)1325 (0.3%) Unknown20 (24.1%)16 577 (4.0%)Surgery No surgery29 (34.9%)31 378 (7.6%)<0.01 Breast-conserving surgery46 (55.4%)302 457 (73.6%) Mastectomy7 (8.4%)74 529 (18.1%) Unknown1 (1.2%)2335 (0.6%)Radiation None/Unknown43 (51.8%)198 607 (48.4%)0.53 Yes40 (48.2%)212 092 (51.6%)Chemotherapy No/Unknown27 (32.5%)227 419 (55.4%)<0.01 Yes56 (67.5%)183 280 (44.6%) Baseline characteristics of patients with PSCCB and IDC Compared with IDC, PSCCB was significantly associated with older age (≥60 years, 60.2% vs 48.5%, p=0.04), male sex (4.8% vs 0.8%, p<0.01), higher pathological grade (III–IV, 74.7% vs 37.9%, p<0.01), higher TNM stage (III–IV, 41.0% vs 14.6%, p<0.01) and a higher rate of distant metastasis (20.5% vs 4.4%, p<0.01). With regard to molecular subtype, PSCCB patients were more likely than breast IDC patients to have triple-negative breast cancer (25.3% vs 7.1%, p<0.01), and PSCCB patients had lower proportions of ER/PR positivity than IDC patients (ER, 14.5% vs 75.6%, p<0.01; PR, 14.5% vs 65.0%, p<0.01). No PSCCB patients had HER2-positive tumours. Regarding treatment options, no significant differences in the administration of radiotherapy were observed between these two groups, but patients with PSCCB were more likely to receive chemotherapy (67.5% vs 44.6%, p=0.03) and less likely to undergo surgery (63.8% vs 91.7%, p<0.01) than those with breast IDC. Comparison of survival between PSCCB and IDC patients The median OS and CSS of PSCCB patients were 23.0 m (95% CI 13.0 to 56.0) and 28.0 m (95% CI 18.0 to 66.0), respectively (figure 1). The 1-, 2- and 5-year OS rates among PSCCB patients were 60.4%, 48.1% and 36.1%, respectively, whereas the 1-, 2- and 5-year CSS rates were 66.6%, 53.0% and 42.4%, respectively. Kaplan-Meier analysis comparing (A) overall survival and (B) cancer-specific survival between patients with PSCCB and IDC in the entire cohort and (C) overall survival and (D) cancer-specific survival between patients with PSCCB and IDC, including only patients with triple-negative breast cancer (TNBC). As shown in figure 1A,B, PSCCB patients had significantly worse OS (HR=5.86, 95%CI 4.78 to 7.91) and CSS (HR=7.58, 95%CI 5.61 to 10.4) than breast IDC patients. As triple-negative breast cancer (TNBC) is associated with a poor prognosis, and 60% of PSCCB patients had TNBC, we further compared CSS and OS between PSCCB and IDC patients who had TNBC. Patients with PSCCB who had TNBC had significantly shorter OS (HR=3.22, 95%CI 1.85 to 5.81) and CSS (HR=3.47, 95%CI 1.93 to 6.62) than those with IDC who had TNBC (p<0.01) (figure 1C,D). In addition, PSCCB patients with luminal A subtype also had a worse prognosis than IDC patients with luminal A subtype (p<0.01). In PSM analysis, 415 patients with IDC were included to match the 83 patients with PSCCB. No significant differences were observed in any of the baseline characteristics between the two groups (online supplemental table 1). In the matched cohort, Kaplan–Meier analysis showed that PSCCB patients had a worse clinical outcome than IDC patients. The 1- and 2-year OS rates among PSCCB patients were 60.4% and 48.1%, respectively, whereas those among IDC patients were 81.7% and 68.1%, respectively (p<0.01). The 1- and 2-year CSS rates among PSCCB patients were 66.6% and 53.0%, respectively, whereas those among IDC patients were 84.7% and 72.7%, respectively (p<0.01) (figure 2). Kaplan-Meier analysis comparing (A) overall survival and (B) cancer-specific survival between patients with PSCCB and IDC in the propensity-score matched cohort and (C) overall survival and (D) cancer-specific survival between patients with PSCCB and IDC in the propensity score-matched cohort including only patients with TNBC. Effect of different treatment regimens on prognosis Next, we analysed the effects of treatment regimens on prognosis among patients with PSCCB. In this cohort, 53 of 83 patients underwent surgery; 46 underwent breast-conserving surgery and 7 underwent mastectomy. Forty (48.2%) and 56 (67.5%) patients received radiotherapy and chemotherapy, respectively. Kaplan-Meier analysis showed that surgery and chemotherapy significantly prolonged OS (surgery: HR=3.65, 95%CI 2.01 to 6.62; chemotherapy: HR=2.67, 95%CI 1.48 to 4.82) and CSS (surgery: HR=3.10, 95%CI 1.60 to 6.00; chemotherapy: HR=2.17, 95%CI 1.10 to 4.26) among patients with PSCCB, whereas radiotherapy had no effect on OS (HR=0.61, 95%CI 0.34 to 1.08) or CSS (HR=0.72, 95%CI 0.38 to 1.34) (online supplemental figure 2A–F). However, there was no significant difference in OS and CSS between patients who underwent breast-conserving surgery and those who underwent mastectomy (OS: p=0.89; CSS: p=0.77). Among the 53 patients who underwent surgery, 27 received a combination of surgery, radiotherapy and chemotherapy; 14 received surgery and chemotherapy; and 10 received surgery alone. Multimodality treatment with surgery, chemotherapy and radiotherapy prolonged OS when compared with surgery alone, but the difference was not statistically significant (online supplemental figure 2G; p=0.09). However, surgery with either chemotherapy or both radiotherapy and chemotherapy significantly prolonged CSS compared with surgery alone (online supplemental figure 2H; p<0.05). In addition, patients with PSCCB who received radiotherapy and chemotherapy had OS and CSS similar to those who underwent surgery alone. Identifying prognostic factors of PSCCB We also conducted univariate and multivariate Cox regression analyses to identify potential prognostic factors for OS and CSS among patients with PSCCB. As shown in online supplemental tables 2 and 3, the univariate Cox analysis showed that age, sex, T stage, M stage, surgery and chemotherapy were significantly associated with OS and CSS. Multivariate Cox analysis after adjusting for confounding factors demonstrated that sex (OS: HR=5.75, 95%CI 1.55 to 21.3; CSS: HR=9.36, 95%CI 2.03 to 43.1), T stage (OS: HR=2.57, 95%CI 1.17 to 5.64; CSS: 95%CI 4.80, 95%CI 1.99 to 11.6) and chemotherapy (OS: HR=0.29, 95%CI 0.13 to 0.65; CSS: 95%CI 0.25, 95%CI 0.10 to 0.64) were independent prognostic factors for both OS and CSS, age was an independent prognostic factor for OS (OS: HR=2.09, 95%CI 1.01 to 4.36). Prognostic nomogram for PSCCB Further, we established OS-/CSS-specific nomogram for PSCCB patients based on the independent prognostic factors from multivariate Cox analysis (figure 3A,B). The C-index for OS-/CSS-specific nomogram was 0.75 (95%CI 0.66 to 0.85) and 0.79 (95%CI 0.69 to 0.89), respectively. The calibration curve indicated that the predicted value was consistent with the actual observation value (figure 3C,D). The ROC analysis of the established nomogram showed that AUC of 1-, 2- and 5-year OS reached 0.85, 0.84 and 0.90, respectively, while AUC of 1-, 2- and 5-year CSS reached 0.85, 0.79 and 0.84, respectively (figure 4A). Moreover, DCA suggested that our nomogram models have a significant positive net benefit from the risk of death and is better than the traditional TNM staging system (figure 4B). Established OS-(A)/CSS-(B)specific nomogram for predicting survival among PSCCB patients. Calibration curve of the nomogram for 1-, 2- and 5-year OS (C) and CSS (D). ROC analysis of the nomogram for 1-, 2-, and 5 -year OS and CSS (A). DCA of the nomograms predicting OS and CSS compared with TNM staging (B). Discussion In 1983, Wade et al reported the first well-documented case of PSCCB.15 However, PSCCB is a rare disease with limited available data; most previous studies have been case reports or case series. Therefore, the clinicopathological characteristics and prognosis of PSCCB remain unclear. In addition, PSCCB is a distinct subtype of breast cancer, and it is unclear whether PSCCB has different features from IDC. We conducted a retrospective study based on 83 patients with PSCCB and 410 699 patients with IDC from the SEER database. We found that PSCCB has distinct clinicopathological features. Compared with IDC, PSCCB was correlated with older age, male sex, higher tumour grade, higher tumour stage, a higher possibility of metastasis, a higher proportion of TNBC and a lower proportion of ER/PR positivity. These features are consistent with those described in previous reports. Although PSCCB typically occurs in women, there have been some reported cases in men.10 Boutrid et al conducted a pooled analysis of 56 PSCCB patients and found that 3.6% (2/56) of the patients were men; in our study, 4 of 83 (4.8%) patients were men. This is much higher than the 0.8% of male IDC patients, suggesting a potential sex predominance of PSCCB. Similarly, previous studies reported that PSCCB is diagnosed in patients aged 25–81 years, with a higher incidence in those over 60 years of age.6 8 16 In our cohort, the affected patients were usually in the sixth decade of life, with an average age at diagnosis of 62.5 years (range, 28–97 years). The proportion of patients older than 60 years was significantly higher than that in IDC patients, suggesting that PSCCB may be more likely to occur in elderly individuals. Because SCC is a subtype of high-grade neuroendocrine carcinoma, PSCCB is highly malignant, with poor pathological differentiation and a higher tumour stage at diagnosis. In our cohort, the proportions of grade III/IV and TNM stage III/IV disease among PSCCB patients were significantly higher than those among IDC patients. In agreement with our results, Boutrid et al also reported a high rate of TNM stage III/IV disease among PSCCB patients (26/56). Similarly, PSCCB patients were more likely to present with lymph node metastases (40.9%) and distant metastases (20.5%) than IDC patients. Previous studies indicated that the lung was the most common metastatic site in PSCCB patients, followed by the bone, liver and other sites. Among the nine cases in our study with adequate information on distant metastases, the lung (n=4), bone (n=4) and liver (n=3) were the most commonly affected anatomical sites. Therefore, routine examination of the lung, bone and other organs to exclude distant metastases is essential for PSCCB patients. Histologically, PSCCB is similar to SCC of the lung, with some evidence of ductal carcinoma in situ with areas of ductal, lobular, and papillary differentiation. Therefore, PSCCB could have similar neuroendocrine markers and hormonal biomarkers, suggesting that PSCCB may have a distinct molecular subtype distribution.9 In our cohort, 35 patients had adequate information on the expression of hormonal markers and HER2, which allowed us to divide patients into those with triple-negative (21/35) and luminal A (14/35) subtypes. The ER/PR-positive rate (14.5%) was significantly lower than that in patients with IDC. With regard to HER2 expression, no HER2 positivity has been reported in previous studies of PSCCB, and no patients in our cohort had HER2-positive tumours. However, information on HER2 expression in the SEER database was only available after 2010, and more than half of PSCCB patients had no accurate information on HER2 expression. Therefore, these results should be confirmed in the future. Previously, PSCCB has been reported to have the highest 5-year OS among all EPSCC cases, which may be associated with the high proportion of patients with early-stage PSCCB.4 13 Our data also showed that tumour stage is correlated with survival; the 5-year OS rates for patients with TNM stages I, II, III and IV were 85.7%, 45.3%, 12.5% and 0%, respectively. It has generally been thought that the prognosis of PSCCB patients is similar to that of patients with small cell lung cancer.12 17 However, our data showed that PSCCB patients have much better prognosis than those with small cell lung cancer, according to the data on 5-year OS rates. The median OS of patients with stage IV PSCCB was only 5 months (95% CI 1.0 to 11.0). Therefore, timely detection seems to be vital, as survival time drastically declines with increased stage. The comparison analysis also showed that PSCCB patients with TNBC and luminal A subtypes had significantly worse clinical outcomes than IDC patients with TNBC and luminal A subtypes, respectively. Because the aggressive features of PSCCB may worsen prognosis, we conducted PSM analysis to adjust for uneven baseline characteristics, and significant differences in OS and CSS were still observed. To the best of our knowledge, there have been no studies on PSCCB using PSM analysis, and this is the first study to compare the prognosis between PSCCB and IDC patients after PSM. Taken together, these results suggest that a worse clinical outcome is an intrinsic characteristic of PSCCB. Moreover, we established a prognostic nomogram to predict the OS and CSS of PSCCB patients. C-index, ROC analysis, and calibration curve showed that the nomogram models have a good discrimination ability. DCA suggested that the nomogram models have potential clinical application value, which is better than traditional TNM staging for predicting OS and CSS. Because PSCCB is rare and has not been extensively investigated, there are numerous treatment strategies, and no recommended optimum therapy protocol has been defined to guide clinicians. Surgery is the main therapy for PSCCB, especially in patients without distant metastases at diagnosis. In our cohort, surgery significantly improved survival, but radiotherapy did not improve survival in patients with PSCCB, which is consistent with the results of previous reports.13 However, chemotherapy did significantly improve prognosis, and multivariate analysis showed that chemotherapy was an independent prognostic factor, suggesting an important role of chemotherapy in PSCCB. Surgery should be supported with chemotherapy and radiotherapy. A previous report showed that systemic chemotherapy was administered to 69.8% of patients, and PSCCB patients treated with surgery and adjuvant chemotherapy had prolonged survival.18 Consistent with this report, our data showed that 67.0% of the patients received chemotherapy. The combination of surgery with chemotherapy prolonged OS and CSS when compared with surgery alone, but the addition of radiotherapy did not improve the prognosis compared with surgery plus chemotherapy. Future research could focus on exploring the optimal strategy for PSCCB patients, for example, the sequence of surgery and chemotherapy, the regimen of chemotherapy and the efficacy of endocrine therapy. Similar to other studies that used the SEER database as a data source, several limitations should be noted, most of which are inherent in a secondary data analysis of this nature.19 First, this retrospective study design limited the collection and analysis of several variables that may be relevant to survival, such as performance status, genetic alterations and chemotherapy regimens. Second, information on some variables was missing. For example, 48 of 83 patients had no information on HER2 status because HER2 information was only available after 2010. This may limit the accuracy of the results. Third, the record pattern of the SEER database may have affected the analysis. For example, for some patients, the records on the use of radiotherapy and chemotherapy stated none or unknown, but they may actually have received these treatments. This bias may have underestimated the treatment effect. Therefore, the repeatability and reliability of these results should be verified in further expanded studies. In conclusion, the present study used a population-based approach to describe the clinicopathological characteristics and survival of PSCCB patients and further compared the survival between PSCCB and IDC using PSM analysis. The data showed that PSCCB has distinct clinicopathological characteristics and PSCCB patients have significantly worse clinical outcomes than IDC patients. And we established the prognostic nomogram for physicians to accurately predict the OS and CSS of PSCCB patients, allowing for individualised clinical decisions for further clinical practice. Supplementary Material Reviewer comments Author's manuscript Footnotes LY and A-HY contributed equally. Contributors: Guarantor: C-CP; Conception and design: LY and C-CP; Administrative support: C-CP; Provision of study materials or patients: LY, A-HY; Collection and assembly of data: LY, A-HY, C-CP; Data analysis and interpretation: LY, A-HY; Manuscript writing: All authors; Final approval of manuscript: All authors. Funding: The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors. Competing interests: None declared. Patient and public involvement: Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research. Provenance and peer review: Not commissioned; externally peer reviewed. Supplemental material: This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise. Data availability statement Data are available upon reasonable request. The raw data downloaded from SEER database. Readers could obtain the data from the corresponding author (Chen-Chen Pu, E-mail: suknight3@126.com) due to reasonable request. Ethics statements Patient consent for publication Not applicable. Ethics approval Not applicable. References Associated Data Supplementary Materials bmjopen-2023-073841supp001.pdf Reviewer comments Author's manuscript Data Availability Statement Data are available upon reasonable request. The raw data downloaded from SEER database. Readers could obtain the data from the corresponding author (Chen-Chen Pu, E-mail: suknight3@126.com) due to reasonable request.