Establishment of the China cervical cancer clinical database
This multicentre retrospective study was approved by the Ethics Committee of the Nanfang Hospital of Southern Medical University (No. NFEC-2017-135) and is registered at the International Clinical Trials Registry Platform Search Port (https://trialsearch.who.int/Trial2.aspx?TrialID=ChiCTR1800017778) under clinical trial registration number CHiCTR1800017778. This retrospective cohort study was conducted following the ethical standards adopted in the 1964 Declaration of Helsinki. The four C database was developed in collaboration with 37 hospitals in mainland China and contained 46,313 cervical cancer patients who received inpatient treatment from 2004 to 2016.
Data collection
Uniformly trained gynaecologists collected the data using standardized data collection and quality control procedures. Patients’ medical records and pathology and examination reports were consulted, and data, including demographics, clinicopathologic features and treatment, were collected, with specific reference to our published studies [10, 11]. Among these parameters, clinical staging was revised according to the 2009 International Federation of Gynaecology and Obstetrics (FIGO) staging standard [12]. After the data collection was completed, two gynaecologists performed independent information verification to ensure accuracy and integrated missing or incomplete data from the supplementary medical records, such as the patient’s case records and the pathology and examination reports.
Follow-up
Because this study was a multicentre retrospective study, trained follow-up personnel at each participating unit performed the follow-up during 1–2 telephone calls. All phone numbers were called uniformly based on the medical record management centre. The follow-up content included survival status, relapse and complications. For patients whose phone number was incorrect or if the patient could not be reached, we used the last visit or report time as the survival time and extracted tumour recurrence from outpatient medical record-related information from the hospital’s outpatient medical records, picture archiving and communication system (PACS) and clinical laboratory information system.
Double data input
Two specially trained gynaecologists input the same data into EpiData software and reviewed the doubtful information to ensure accuracy.
Data storage
After collecting all case information and follow-up data and completing double-input verification, the data were aggregated and managed by a professional to establish a unified database.
Inclusion and exclusion criteria
The following inclusion criteria were used: (1) FIGO stage IB1-IIA2 (FIGO 2009 staging system); (2) age ≥ 18 years old; (3) biopsy or postoperative pathology confirmed as squamous cell carcinoma, adenocarcinoma or adenosquamous carcinoma; (4) no neoadjuvant chemotherapy or radiotherapy before surgery; (5) Q-M type B or type C abdominal radical hysterectomy (ARH) + PL ± PAL; and (6) complete postoperative pathological data. The following exclusion criteria were used:(1) pregnancy with cervical cancer; (2) cervical stump cancer; (3) combined with other malignancies; (4) patient was lost to follow-up; and (5) patient did not meet the inclusion criteria. The scope of PAL includes PALN biopsy, low abdominal PALN resection, and high abdominal PALN resection.
Case-control matching
The factors included in the multivariate analysis were age, FIGO stage, whether a PALN was resected, histological type, vaginal margin, parametrial infiltration, tumour diameter, deep stromal invasion, lymphovascular invasion (LVSI), and whether the postoperative adjuvant treatment was standard. Whether the postoperative adjuvant treatment was standard was based on pathological factors according to the guidelines for treatment [2, 13]: one or more postoperative pathological high-risk factors (positive lymph nodes, parametrial infiltration or positive margins): external-beam radiation therapy + platinum-containing concurrent chemotherapy (level of evidence 1) ± vaginal brachytherapy; and intermediate-risk factors (tumour size, deep stromal invasion, LVSI) according to the “Sedlis criteria” (level of evidence 1): external-beam radiation therapy ± concurrent platinum-containing chemotherapy (simultaneous chemotherapy evidence level 2B). The definition of “inadequate” in the classification of postoperative adjuvant treatment was when there are one or more postoperative high-risk factors, postoperative adjuvant therapy would be chemotherapy only or no treatment or when there are two or more intermediate-risk factors after surgery, the postoperative adjuvant therapy would be chemotherapy only or no treatment. The definition of “over” in the classification of postoperative adjuvant treatment was when there are no postoperative high-risk factors and intermediate-risk factors or when there are no postoperative high-risk factors and only one postoperative intermediate-risk factor, the postoperative adjuvant treatment would included postoperative radiotherapy or chemoradiotherapy or chemotherapy.
Because the clinicopathological data of the PALN unresected and resected groups may have differed, we used propensity score matching (PSM)/case-control matching to balance the factors that were different between the two groups to ensure that the groups were comparable.
Outcome evaluation
The main observation outcomes were the 5-year overall survival (OS) rate and the 5-year disease-free survival (DFS) rate between the PALN unresected and resected groups of the overall and different PLN metastasis states. OS was defined as the date of diagnosis to death from any cause or the last effective follow-up. DFS was defined as the date of diagnosis to death, relapse or the last effective follow-up.
Statistical analysis
Data analysis was performed using SPSS statistical software (version 23.0, SPSS Inc., Chicago, IL, USA). Two independent sample t tests were used for continuous variables, and the X2 test or nonparametric test was used for categorical variables or grade variables. The log-rank test in the Kaplan-Meier (KM) method was used to compare the 5-year survival outcomes (OS, DFS) of the two groups. The Cox proportional hazards regression model was used to calculate hazard ratios (HRs) and 95% confidence intervals (CIs) for the multivariate analysis. P < 0.05 was considered statistically significant. Statistical experts reviewed all statistical methods and statistical processes in this study.