The Nomograms for Predicting Cancer-specific Survival in Patients With Ovarian Cancer After Surgery


 Objectives: Clear cell adenocarcinoma of the ovary (CCAO) and mixed cell adenocarcinoma of the ovary (MACO) were one of the gynecological malignancies.Methods: Univariate and multivariate cox regression analysis were used to determine prognostic factors. Drawing nomograms, the receiver operating characteristic (ROC) curve and the calibration curve was applied to evaluate the agreement of the nomogram. The survival analysis was constructed to the high-risk factors.Results: The nomogram was constructed and had a better discrimination. The calibration curves indicated that the nomograms had good calibration capabilities.4 or more regional lymph nodes removed by surgery was beneficial to the patient's prognosis. Conclusions: Our study analyzed the prognosis of CCAO or MACO patients, and constructed a predictive nomogram with good accuracy.


Introduction
Ovarian clear cell carcinoma of the ovary (CCCO) is a rare histologic subtype of ovarian cancer. It is often misdiagnosed and missed due to the lack of typical clinical manifestations and speci c diagnostic methods. Simultaneously, the clinical prognosis is poor because it is too late to nd out. At present, surgical treatment is the main clinical treatment, supplemented by comprehensive treatment of chemotherapy and neoadjuvant chemotherapy [1][2][3]. Patients with CCCO should be treated strictly by AJCC staging surgery [4]. Tumor reduction surgery was positively correlated with postoperative survival, while the size of residual tumor lesions was negatively correlated with survival rate [5]. However, there is currently a lack of clinically established prognostic models for patients with CCCO after surgical treatment, and the impact of postoperative lymph node dissection on the prognosis has not been established.
This study used SEER*Stat software to retrospectively analyze the National Cancer Institute database

Data Collection
The SEER database is a public database that collects information on millions of patients with malignant tumors in some states and counties of the United States, including the incidence Status, treatment status, prognosis and death. In this study, a total of 1127 cases for CCAO and 1064 cases for MCAO were selected for retrospective analysis using SEER*Stat software. The inclusion and exclusion criteria were as follows. Inclusion criteria: (1) All included are primary ovarian tumors, (2) The pathological tissue ICD-O-3 classi cation were 8130 (CCAO) and 8323 (MCAO), (3) The diagnosis period was 2010 to 2015 years, (4) All patients were treated with surgery. Exclusion criteria: (1) Incomplete information such as tumor differentiation and staging information and corresponding variable indicators; (2) Incomplete follow-up information; (3) Carcinoma in situ; (4) The patients died within 1 month.

Statistical analysis
The in uencing factors of CCAO or MCAO were selected in this study including age, race, laterality, grade, the seventh edition of American Joint Committee on Cancer (AJCC) staging for the extent of tumor (T), extent of spread to lymph nodes (N), and presence of metastasis (M), regional lymph nodes surgery, CSS rate, survival status. The grade included the following types: well-differentiated (G1), moderatelydifferentiated (G2), poorly-differentiated (G3), and undifferentiated (G4). The cox proportional hazard regression model was used for univariate and multivariate analysis, and the independent factors affecting the prognosis of CCAO or MCAO were obtained. Drawing a nomogram to predict 3-year and 5year CSS rate, using receiver operating characteristic (ROC) curve to evaluate the predictive ability of the model. Calibration curve were respectively adopted to evaluate the relationship between the actual results of the model and the expected results probability of association in the nomogram. The Kaplan-Meier plotter analysis was to explore the CSS rate of the risk factors. All data analyses were conducted using R software (version 3.6.0; http://www.r-project.org/). P value of less than 0.05 was considered to statistical signi cance.

Cox regression analysis results
For CCAO patients, we screened out seven independent prognostic factors by univariate and multivariate cox regression analysis. Age, race, laterality, AJCC T, AJCC N, AJCC M and regional lymph nodes surgery were closely related to CSS in CCAO patients (Table 2). For MCAO patients, we obtained ve independent prognostic factors by the above methods, such as grade, AJCC T, AJCC N, AJCC M, regional lymph nodes surgery (Table 3).

Nomogram construction
The above independent in uencing factors were screened out and assigned, and the maximum amount of change was assigned as 100 points, and then the scores corresponding to the individual indicators of each research object were obtained. Finally, the total score was obtained by summarizing the scores of each of the selected variables. The total score could be used to nd the 3-year and 5-year CSS. Two Nomograms predicting CSS for CCAO (Fig. 1A) or MCAO ( Fig. 1B) patients were established as follows.

Discussion
CCCO was a type of epithelial ovarian cancer, which was highly malignant and had a poor prognosis [6].
Studies showed that the incidence of Asians was signi cantly higher than that of other races [7,8]. The median age of this disease was 55 years old [9]. In our study, both CCAO and MCAO patients were selected for surgical treatment, whose age were between 50-59 years old (CCAO: 34.69%, MCAO: 32.05%), and most of them were white people. For CCAO patients, who younger than 40 years old were apt to own a poor prognosis, while there is no statistical difference between the age and poor prognosis for MCAO patients. Interestingly, bilateral ovarian disease of the patients with CCAO increased their poor prognosis. As the level of AJCC T staging increasing of CCAO or MCAO patients, the risk of adverse prognosis improved. Lymph node metastasis and distant organ metastasis were the important factors for its adverse outcome. Importantly, Multi-regional lymph node surgical removal (4 or more regional) could improve a good survival outcome for CCAO patient. In MCAO patients, the worse of the grade, the worse of the survival prognosis. Regional lymph node dissection could improve their poor prognosis. Many studies have shown that lymphadenectomy was very controversial for their prognosis [10][11][12], but we recommend that patients with CCAO or MCAO could undergo lymphadenectomy during surgery.
The nomogram model is currently an important tool for assessing the staging system and prognosis, and it is widely used in clinical research [13][14][15]. Our research comprehensively analyzed the above independent factors by establishing nomogram prediction model, using to predict the 3-year and 5-year CSS in CCAO or MCAO patients. In the nomogram, we could quickly screen out the independent risk factors. And according to their total scores added, the 3-year and 5-year CSS of patients with CCAO or MCAO after surgery were displayed, which had important clinical value.
At the same time, the ROC curve was established for the two models, which showed that the two models had better discrimination and better diagnostic effects. By drawing the calibration curve, we found that the predicted value and the actual value in the curve diagram of the two prediction models were consistent, which indicated that the prediction models had better calibration capabilities. Therefore, the establishment of the two models was satisfactory and could be used for clinical preliminary treatment evaluation. Some studies have shown that for CCCO patients, the survival rate after surgery was the key to evaluating treatment[16-18]. Our survival analysis showed that the survival time of patients with highrisk factors was shorter than that of patients with low-risk factors.
The shortcomings of this study were that the prognosis of patients was evaluated only from the perspective of surgery, and the included indicators had some limitations. At the same time, the study did not conduct external veri cation. Therefore, it is necessary to collect more clinical data to verify this and provide a more favorable guarantee for the implementation of precise and individualized treatment. The survival curves of high-risk and low-risk score in CCOA (A) and MCAO (B) patients