Prognostic factors and treatment comparison in small cell neuroendocrine carcinoma of the uterine cervix based on population analyses

Abstract Objective We aimed to assess the impact of the treatment modality on the outcome of small cell neuroendocrine cervical carcinoma (SCNEC) using the Surveillance Epidemiology and End Results (SEER) database. Methods Patients from the SEER program between 1981 and 2014 were identified. Significant factors for cancer‐specific survival (CSS) and overall survival (OS) were analyzed using the Kaplan‐Meier survival and Cox regression methods. Results A total of 503 SCNEC patients were identified. The 5‐year CSS and OS were 36.6% and 30.6%, respectively. The International Federation of Gynecology and Obstetrics (FIGO) stage I to IV distributions was 189 (37.6%), 108 (21.5%), 95 (18.9%), and 111 patients (22.0%), respectively. Within the patients with known treatment strategies, 177 (45.9%) were treated with radical surgery and 209 (54.1%) underwent primary radiotherapy. Local treatment strategies were independent prognostic factor for CSS and OS. The 5‐year CSS for radical surgery and primary radiotherapy was 50.0% and 27.9%, respectively (P < .001). The 5‐year OS for those who received radical surgery and primary radiotherapy was 57.8%, and 29.6%, respectively (P < .001). In FIGO stage I SCNEC, patients treated with radical surgery had superior CSS (P = .001) and OS (P = .003) than those with primary radiotherapy. However, in FIGO stage II and III SCNEC, there were no differences in CSS and OS with respect to different local treatment strategies. Our results also found that the addition of brachytherapy impacted OS in the FIGO stage III SENCE (P = .002). The 5‐year CSS and OS of patients with FIGO IV were only 11.7% and 7.1%, respectively. Conclusions SCNEC is a rare disease with aggressive clinical behavior. The findings indicate that radical surgery should be suggested for early‐stage SCNEC and combining radiation therapy with brachytherapy should be suitable for patients with advanced stage.


| INTRODUCTION
Nearly 2 to 5% of all cervical carcinomas are small cell neuroendocrine uterine carcinomas (SCNEC). [1][2][3] Compared with common histological types, SCNEC seems to be highly aggressive and has a worse prognosis even in its early stages. [4][5][6][7][8][9] Because of the rarity of the disease, most of the previous studies were clinical case-reports or limited series and from single institutions. To date, there are no published guidelines for the standard local therapy for SCNEC. Prognostic factors including age, tumor size, tumor stage, metastases of the lymph node, and margin status were examined with various results. The 5-year survival of early-stage tumors was 30%-46%, and only 0%-15% in advanced-stage carcinoma. 10 The therapeutic approach for SCNEC still remains a challenge.
Given the aggressive clinical behavior of the disease, to improve treatment outcome a potential treatment strategy different from that of common histological types is required. Therefore, the aim of our research was to estimate the effect of treatment modality on survival in SCNCE using the Surveillance Epidemiology and End Results (SEER) database.

| Data source
Our research data were extracted from the SEER database maintained by the National Cancer Institute. The SEER database is composed of 18 population-based cancer registries and covers nearly 27.8% of the US population (based on the 2010 Census). We accessed data files with the SEER ID 13027-Nov2018 and identified patients with a primary diagnosis of SCENC between 1981 and 2014. The primary site was used for the pathological diagnosis of the illness according to the third edition of the International Oncology Classification (ICD-O-3). We identified local treatment strategies using codes for surgery and radiotherapy (RT). Radical hysterectomy, modified radical hysterectomy, complete hysterectomy, and pelvic exenteration were included in curative surgery. External beam radiation therapy (EBRT), with or without brachytherapy before curative surgery, was defined as local radiation treatment.

| Clinicopathological factors
The clinicopathological and demographic variables were gathered as follows: age of diagnosis, race, tumor grade, International Federation of Gynecology and Obstetrics (FIGO) stage, tumor size, state of lymph node, and treatment strategies including radical surgery and primary RT. Duration of follow-up and vital status, including the cause of death was also included. The primary endpoints were CCS and OS.

| Statistical analysis
We used univariate and multivariate Cox proportional hazards models to estimate prognostic factors for CSS and OS. Factors that were significantly related to CSS and OS in univariate analyses were included in the multivariate analysis. Calculation of CSS and OS were evaluated using the Kaplan-Meier survival and Cox regression proportional hazard methods. All calculations were carried out with the SEER-Stat software (SEER*Stat 8.3.5). Two-sided P-values were calculated and a cutoff of P < .05 was statistically significant.

| Patient characteristics and treatment
A total of 503 SENCE patients were reported in the SEER database from 1981 to 2014. The demographic, clinicopathological, and treatment characteristics of the population are listed in Table 1

| Survival
The 3-and 5-year CSS were 40.9% and 36.6%, respectively ( Figure 1A) and the 3-and 5-year OS were 36.2% and 30.6%, respectively ( Figure 1B). The 5-year CSS in patients treated with radical surgery and primary RT was 50.0%, and 27.9%, respectively (P < .001; Figure 2A). The 5-year OS according to radical surgery and primary RT was 57.8%, and 29.6%, respectively (P < .001; Figure 2B). The outcome of the local treatment strategy was different from the FIGO stage. Patients treated with radical surgery had significantly improved CSS (P = .001; Figure 3A) and OS (P = .003; Figure 3B) compare to those with primary RT in stage I SCNEC. For SCNEC patients with stage II, however, there were no differences in CSS (P = .67; Figure 4A) and OS (P = .64; Figure 4B), according to different local treatment strategies. There were also no differences in CSS (P = .18; Figure 5A) and OS (P = .42; Figure 5B) between the patients with stage III according to different local treatment strategies.
We also investigated the impact of different RT methods on the survival of primary RT-treated patients and found that RT combined with brachytherapy had an impact on both CSS (P < .001; Figure 6A) and OS (P = .002; Figure 6B) in stage III SENCE.

| DISCUSSION
This study demonstrated poor CSS and OS for all stages of disease illustrating the aggressive nature of SCNEC. We analyzed the impact of local therapeutic approaches on survival of SCNEC patients from the SEER registry. In our study, the patients with early stages preferred radical surgery and those with advanced stages preferred primary RT. Based on our results, curative surgery would be recommendable as the optimal local therapeutic approach for early-stage SCNEC whereas combining RT with brachytherapy should be suitable for patients with advanced stage. Currently, there are no prospective studies comparing prognosis in SCNEC patients with different local therapeutic approaches because of the rarity of SCNEC. There were, however, some reports comparing the survival of SCNEC patients treated with radical surgery versus primary RT which showed different results. In a multicenter study, Chen et al. 11 observed that in stage I-II patients, primary RT with aggressive chemotherapy had superior 5-year OS than surgery (78% vs 40%). In contrast, a systematic review and meta-analysis concluded that there was no significant difference in the survival outcomes with or without adjuvant RT in SCNEC. 12 Furthermore, Cohen et al. 13  However, the results might be affected by the broad definition of cancer-directed surgery which included biopsy, ablation, and local excision. In our study, age (≤50 vs >50 years) was found to be an independent factor affecting OS (HR = 1.561, P = .034).  Several reports have also found a worse survival outcomes with increasing age. 5,7,14 In a retrospective study of 130 patients with SCNEC, age older than 60 years was proved to be an independent prognostic factors for CSS regardless of stage. 7 Among the patients with IIB-IVB stage, those aged younger than 45 years had significantly lower risk of cancer death than those aged 45 to 60 years (HR, 3.4; P = .035). Zhou et al. 5 who analyzed data of 208 patients with stage I-II SCNEC found that age was an independent prognostic factor for OS (HR = 1.017, P = .004). Similarly, Hou et al. 14 found a significant OS benefit in younger cases (P = .005) with high-grade neuroendocrine carcinoma in a retrospective analysis.
The more advanced stages were found to be correlated with worse survival in several studies. [13][14][15][16][17][18][19] Zhang et al. 15 showed that the median survival time (MST) of SCNEC patients with early-stage (I-IIA) was longer than those with advanced-stage (IIB-IV) (60 vs 30 months, P = .016). However, the FIGO stage was not an independent prognostic factor in the multivariate analyses. Another study of the Taiwanese multi-institutional database concluded that FIGO stage was significantly related to failure-free survival (FFS) (HR: 2.28, P = .001) and CSS (HR: 2.27, P < .001) in multivariate analyses. 17 Similarly, in our study, multivariate analyses showed that FIGO stage was an independent factor affecting CSS and OS (CSS: HR = 1.414, P = .002; OS: HR = 1.356, P = .005).

F I G U R E 4 Cancer-specific survival (A) and overall survival (B) by radical surgery versus primary RT in FIGO stage II patients
Lymph node metastasis probability is reported ranging from 34.9% to 65% in SCENC in several studies, 11,17,18 which is higher than that reported in common histological types of cervical cancer. 20 To date, the prognostic value of lymph node status in SCENC is controversial. In this study, lymph node metastasis occurred in 51.2% of SCENC patients, and the status of lymph nodes was proved to be an independent prognostic factor in multivariate analysis. The results were consistent with the research by Wang et al. 17 in which node metastasis was also selected as an independent variable. Nonetheless, several studies failed to demonstrate the predictive value of lymph node status on survival. 11,13,18 More studies are warranted to assess the prognostic value of lymph node status in SCENC patients.
Due to the similar biological behavior with small-cell lung cancer, SCENC was usually treated in a similar fashion as SCLC. [21][22][23] Several studies have recommended adjuvant chemotherapy or concurrent chemotherapy for SCENC patients according to their results. 11,13,24,25 Several studies reported that, after 1990, most patients underwent chemotherapy-dominant comprehensive therapy. 17,[24][25][26] In our study, the chemotherapy was significantly related to favorable CSS and OS in the univariate analyses, however it was not an independent prognostic factor in multivariate analysis. Given that the information about chemotherapy regimens was missing in the SEER database, we were unable to analyze the effect of different chemotherapy regimens on the survival of SCENC patients in this study. The pathological factors, including parametrial invasion, surgery margin, and lymphovascular invasion are significantly related to survival in cervical cancer. The SEER database did not, however, provide information on the abovementioned pathologic factors. It remains controversial whether the pathological factors in SCENC patients have prognostic value. Wang et al. reported that positive surgical margin was a significant prognostic factor for FFS but not for CSS; parametrial extension and lymphovascular invasion had no prognostic value in multivariate analyses. 17 On the contrary, Chen failed to identify the surgical margin as a prognostic factor. 11 Parametrial involvement and lymphovascular invasion were also reported to have no prognostic value in other studies. 11,13,18 Therefore, the prognostic factors in SCENC are different from those of common histological types of cervical cancer.
Brachytherapy acted an important part in the definitive management of common histological types of the uterine cervix. 27,28 In this study, EBRT plus brachytherapy improved CSS and OS in patients with stage III disease. However, the pelvic control could not be assessed together with the brachytherapy because the endpoint was unavailable in the current database. Whether the addition of brachytherapy could improve pelvic control for SCNEC patients remains controversial. Large database studies with detailed information about disease relapse are required to answer these questions.
Due to its methodology and structure, this study has limitations. First, the information regarding chemotherapy regimens and the sequence with surgery or RT were not available in the SEER database, which may have affected the assessment of the prognostic value of the therapeutic approaches. Second, the SEER database did not provide specific information on clinical or surgical staging hence the conclusions could be affected by staging bias. Third, given that the patients with early stages preferred surgery and patients with advanced stages preferred radiation, the survival outcomes associated with the different modalities would be affected. Lastly, we were unable to exclude the use of palliative RT because there was no information regarding the dose of radiation or the size of the radiation fields in the SEER database.
In conclusion, SCNEC is a rare disease with highly aggressive features and poor survival. This study suggests that surgery would be the optimal local therapeutic approach for early-stage SCNEC. For patients with advanced stage, combining primary RT with brachytherapy seems to be the better treatment. We hope that our study contributes to the foundation of knowledge regarding this rare and aggressive disease and inspires more prospective studies to help define optimal local management for SCNEC.