Clinicopathological characteristics and prognostic factors of cervical adenocarcinoma

We aimed to assess the clinicopathological features and to determine the prognostic factors of cervical adenocarcinoma (AC). Relevant data were extracted from surveillance, epidemiology and end results database from 2004 to 2015. The log-rank test and Cox proportional hazard analysis were subsequently utilized to identify independent prognostic factors. A total of 3102 patients were identified. The enrolled patients were characterized by higher proportion of early FIGO stage (stage I: 65.9%; stage II: 14.1%), low pathological grade (grade I/II: 49.1%) and tumor size ≤ 4 cm (46.8%). The 5- and 10-year cancer-specific survival rates of these patients were 74.47% and 70.00%, respectively. Meanwhile, the 5- and 10-year overall survival (OS) rates were 71.52% and 65.17%, respectively. Multivariate analysis revealed that married status, surgery as well as chemotherapy were independent favorable prognostic indicators. Additionally, aged > 45, tumor grade III/IV, tumor size > 4 cm, advanced FIGO stage and pelvic lymph node metastasis (LNM) were unfavorable prognostic factors (all P < 0.01). Stratified analysis found that patients without surgery could significantly benefit from chemotherapy and radiotherapy. In addition, chemotherapy could significantly improve the survival in stage II–IV patients and radiotherapy could only improve the survival in stage III patients (all P < 0.01). Marital status, age, grade, tumor size, FIGO stage, surgery, pelvic LNM and chemotherapy were significantly associated with the prognosis of cervical AC.

Study population. SEER*State v8.3.6 (released on August 8, 2019) was utilized to select and identify qualified subjects, which includes 18 SEER regions from 2004 to 2015 (2018 submission). The inclusion criteria were as follows: (1) primary cervical AC patients; (2) the diagnosis of cervical AC was based on ICD-O-3; coded as 8140-8490 19,20 . Patients were eliminated if they had: (1) more than one primary malignancies; (2)reported diagnosis source from autopsy or death certificate or without pathological diagnosis; (3)without certain necessary clinicopathological data, including surgical style as well as FIGO stage; (4) without prognostic information. The rest of subjects were enrolled as the initial cohort of SEER.
Covariates and endpoint. The following clinicopathological parameters were analyzed: year of diagnosis (2004-2007, 2008-2011, 2012-2015) 21 ; marital status (unmarried, married);race (black, whiteor others);insured status (uninsured/unknown, any medicaid/insured); age(≤ 45, > 45); grade (grade I/II, grade III/IV, unknown); FIGO stage (stage I, II, III, IV); tumor size (≤ 4 cm, > 4 cm, unknown); pelvic lymph node (LN) dissections (none or biopsy, removal of 1 to 3 regional LNs, removal of ≥ 4 regional LNs), pelvic lymph node metastasis (LNM) (positive, negative and unknown);surgery (no surgery, local tumor excision, total hysterectomy), chemotherapy (no/unknown, yes) and radiotherapy(no/unknown, yes). Patients with widowed or single (never married or having a domestic partner) or divorced or separated status were all classified as unmarried 22,23 . All of the eligible cases were re-identified according to the 2018 FIGO staging criteria 24,25 . Median age at diagnosis was 45 years old in our study, which was used as the cutoff value for age classification. Meanwhile, the classification of tumor size and age was also based on previous researches 6,26 . CCRT was defined as the addition of chemotherapy during radiotherapy. Definitive radiotherapy indicated that only radiotherapy was used in the treatment 27 . The endpoints of our research included overall survival (OS) and cancer-specific survival (CSS). The former was defined as the duration from diagnosis to all-cause death, and the latter referred to the duration from diagnosis to cervical AC-caused death.
Statistical analyses. Kaplan-Meier (K-M) method was employed to estimate the univariate analysis, followed by log-rank test for assessing the differences of CSS and OS among different groups. Variables with P values ≤ 0.1 in the univariate analysis were further incorporated into the multivariate Cox proportional hazard analysis. In addition, stratified analysis was performed by using Cox regression analysis. SPSS software (SPSS Inc., Chicago, USA, version 19.0) was utilized for statistical analysis, and GraphPad Prism 5 was utilized for plotting survival curves. These softwares have recieved permission and freely available. A two-sided P < 0.05 was considered as statistically significant. These softwares have been approved.

Stratified analysis of the effect of chemotherapy and radiotherapy on survival.
To explore the benefits of chemotherapy and radiotherapy, we performed stratified analysis of patients with different FIGO stage and surgical style. As a result, patients with stage III/IV could significantly benefit from chemotherapy (both CSS and OS) (P < 0.001), and stage II patients could benefit in terms of OS (P = 0.004). Meanwhile, patients without surgery could also significantly benefit from chemotherapy and radiotherapy (P < 0.05). In addition, only patients with stage III could significantly benefit from radiotherapy (P < 0.001) ( Tables 4, 5).

Discussion
This population-based study revealed the clinicopathological features as well as survival of patients with cervical AC. Cervical AC accounts for only approximately 20-25% of all cervical carcinomas 2, 3 . AC is the second most common type of primary cervical cancer, secondly only to SCC 28 . Previous studies predominantly enrolling patients with SCC have provided most of the present therapeutic knowledge on cervical cancer 29,30 . However, the different outcomes of AC have been rarely reported. Furthermore, prospective studies have not solely focused on the treatment of AC. Consequently, our understanding of the natural history, prognosis factors and optimal management of cervical AC is limited 31    www.nature.com/scientificreports/ involvement were independent prognostic factors for disease-free survival (DFS) and OS 13 . In addition, the analysis of 46 patients with stage I-IV cervical AC revealed that FIGO stage was the only independent prognostic factor for both DFS and OS 11 . A retrospective Dutch study assessing 305 cases of cervical AC found that tumor size, tumor grade and LNM remained as significant independent predictors for survival 12 . Although most of these studies are small-size and single-center retrospective studies, with consistent results to ours. In addition, we also found that marital status is an independent prognostic factor for cervical AC. The same therapeutic strategy is recommended for SCC and AC according to the present guidelines. Nevertheless, there have been no consistent data concerning the therapeutic efficacy in different histological classification 7 . Surgery and radiotherapy are recommended as the primary therapeutic regimes for early-stage cervical cancer in accordance with NCCN guidelines 8 . In addition, the 5-year OS rates for stage IA1 and stage IA2 lesions were 96.5% and 99.4%, respectively, for radical hysterectomy, 96.6% and 100%, respectively, for local excision, 98.4% and 96.9%, respectively, for simple hysterectomy in a study enrolling 1567 patients with cervical AC 32 . Our study also found that surgery is an independent favorable prognostic factor.
Radiotherapy is an alternative option for patients who are not suitable for surgery or who refuse surgery. For patients with stage IB2-IVA cervical cancer, concurrent cisplatin based-chemoradiotherapy plus brachytherapy was the standard therapeutic regimen 7 . Our study found that radiotherapy and chemotherapy could provide significant survival benefits among patients without surgery. However, in terms of tumor stage, only patients with stage III could gain significant survival benefits from radiotherapy. The worse efficacy of cervical AC is possibly caused by insensitivity of radiotherapy. Cervical AC patients have been reported to have poorer complete response (CR) as well as local control rates, therefore requiring longer time to obtain CR than SCC populations following CCRT or definitive radiotherapy 29,33,34 . In addition to pathological type, tumor size and the type of human papilloma virus(HPV) infection were also considered to be important causes for the radiosensitivity of cervical cancer 35,36 .
In consideration of the poor outcomes of patients with cervical AC, more effective protocols are required for these patients. Adjuvant chemotherapy or neoadjuvant is a possible strategy. According to a Chinese clinical trial, 880 patients with FIGO stage IIB-IVA cervical AC were randomly assigned to receive only CCRT or CCRT with one cycle of neoadjuvant chemotherapy and two cycles of consolidation chemotherapy. Subsequently, patients treated by CCRT along with chemotherapy had better OS, DFS and local control after a median follow-up of 60 months. The above outcomes implicate that combined CCRT and chemotherapy is promising to enhance the survival of patients with cervical AC 37 .  www.nature.com/scientificreports/ The NCI-supported SEER database is the most authoritative and largest source for tumor incidence and survival. The large-scale, publicly available SEER dataset can be reliably used to guide anti-cervical AC therapy. As far as we know, our research includes the largest subjects to investigate prognostic parameters for cervical AC in the past 10 years. Inevitably, there are still several limitations in our study. Firstly, selection bias and the effects of inaccessible variables from the SEER dataset are unavoidable due to the nonrandomized nature of our research 17,38 . Secondly, information on HPV 7, 39 , comorbidities and medication use were inaccessible from SEER database, which are considered as valuable indicators for survival of cervical cancer. Thirdly, SEER fails to provide all data to completely address our hypothesis, such as detailed information on chemotherapy and radiotherapy. Nevertheless, the currently accessible information from SEER database could fit our objectives. While the above-mentioned issues should be further addressed.

Conclusions
Marital status, age, grade, tumor size, FIGO stage, pelvic LNM, surgery and chemotherapy were significantly associated with the prognosis of cervical AC. Patients without surgery could significantly benefit from chemotherapy and radiotherapy. Stage II-IV patients could significantly benefit from chemotherapy. In addition, only stage III patients could obtain significant survival benefit from radiotherapy. This is the largest study to investigate the clinicopathological characteristics and outcomes for patients with cervical AC. The present findings in our study are vital to the disease management and future prospective studies for this rare cancer.