Article Text
Abstract
Background In 2018 the International Federation of Gynecology and Obstetrics (FIGO) revised the staging system of cervical cancer. This study aimed to assess the quality of staging early cervical cancer in China before the revision.
Methods This multicenter retrospective study included 34 tertiary hospitals in China. Medical records of patients with cervical cancer who underwent primary surgical treatment between January 2010 and December 2015 were reviewed retrospectively. All patients were clinically staged according to the 2009 FIGO staging system. Eligibility criteria included: histopathologically confirmed cervical cancer; 2009 FIGO stage IA–IIA2 based on 2009 FIGO staging system; primary surgical treatment including extrafascial, type II or type III radical hysterectomy; radical trachelectomy; with or without pelvic lymphadenectomy; regardless of surgical route via laparotomy or laparoscopy; and complete clinical and pathological data. Patients who received non-surgical treatment, neoadjuvant treatment, or those with incomplete data were excluded. The accuracy of clinical staging was assessed by comparison between clinical and pathologic stage using the latter as the reference standard.
Results A total of 23 933 cases of cervical cancer were identified and 12 681 fulfilled the inclusion criteria. Of these patients, 69.6% were staged accurately, 9.4% were clinically understaged, and 21.0% were clinically overstaged. The accuracy of stage IA, IB1, IB2, IIA1, and IIA2 was 90.0%, 87.5%, 57.4%, 20.3%, and 25.5%, respectively. The causes of stage inaccuracy were as follows: vaginal involvement (62.3%), maximal tumor diameter (24.6%), extent of cervical stromal invasion (7.1%), parametrial invasion (5.8%), bladder or rectal infiltration (0.1%), and distant metastases (0.1%).
Conclusion The accuracy of staging early cervical cancer in China was suboptimal before the revision of the staging system, especially for IIA1 and IIA2. The most common reasons for staging inaccuracy were vaginal involvement and tumor diameter.
- cervical cancer
- uterine cervical neoplasm
- staging
- quality
- early stage
Statistics from Altmetric.com
HIGHLIGHTS
The quality of staging early cervical cancer in China before the recent FIGO staging revision was suboptimal
Accuracy of stage IIA1 and IIA2 were 20.3% and 25.5%, respectively
Vaginal involvement and tumor diameter are the main contributors to staging inaccuracy
Introduction
Cancer staging, which reflects the anatomical extent of a cancer, is essential for patient care and cancer research.1 Cancer staging can be categorized as clinical staging and pathologic staging. The pathologic stage is only eligible to patients who have undergone surgical treatment and is based on information obtained from both the surgical procedure and pathologic examination of the surgical specimen. Unlike the pathologic stage, the clinical stage is applicable to all patients and is based on information provided by physical examination, imaging, or endoscopic methods before definitive treatment.1 2
Clinical staging has been developed toward two different roadmaps. One focuses on the accuracy of staging and prediction of prognosis. Patients are staged by a multidisciplinary process with the combination of physical examination, imaging modalities, or invasive staging procedures.3 The other emphasizes the comparability of staging between different cancer centers. Patients are staged by generally available staging procedures across the world. For instance, in cervical cancer the International Federation of Gynecology and Obstetrics (FIGO) has traditionally limited the staging procedures to pelvic examination, with the addition of plain imaging tests such as intravenous pyelography, barium enema, or X-ray of chest and bone.4
This previous principle of clinical staging in cervical cancer has been challenged over the past decades. According to the studies of practice pattern, advanced imaging tests mainly refer to cross-sectional imaging techniques such as magnetic resonance imaging (MRI) or computed tomography (CT) and functional imaging techniques such as positron emission tomography (PET). These have been increasingly used in the clinical staging of cervical cancer in high-income countries.5 6 Therefore, even if this principle of clinical staging is adhered to in middle- and low-income countries, the comparability in different institutions across the world can still not be guaranteed. Second, according to epidemiologic studies, a shift toward increased diagnosis of early stage disease has been observed in many countries.7 8 In the early stage, the imaging tests allowed by the FIGO classification have been reported to be of low efficacy.9 10 Third, with the growing demand for fertility preservation and concerns about quality of life, the surgical techniques have become diversified and sophisticated.11 In the new FIGO cancer report 2018, imaging and pathologic findings, where available, can be used to assign the stage.12 This revision in the staging system is expected to decrease classification error of clinical staging.
The aim of this multicenter retrospective study was to assess the quality of staging early cervical cancer in China before the recent revision of the staging system. The overall accuracy of clinical staging in our study could provide historical data for evaluating the impact of the revision. The analysis of the causes of staging inaccuracy could also facilitate prediction of the revision and further optimization of clinical staging.
Methods
In 2014 a large-scale retrospective multicenter study including 34 tertiary hospitals was launched in China. There were 24 general hospitals, 5 cancer hospitals, and 5 maternity hospitals.13 After approval by the Institutional Review Board of Nanfang Hospital, Southern Medical University, People’s Republic of China and all other respective participating hospitals, medical records of 23 933 patients treated between January 2010 and December 2015 were reviewed. All data including demographic characteristics, clinical stage, and surgicopathologic findings were collected and stored in a database using standardized data entry and quality control procedures. Patients were anonymized before being entered in the database and the information was kept confidential.
The eligibility criteria for the current study included: histopathologically confirmed cervical cancer; 2009 FIGO stage IA–IIA2; primary surgical treatment including extrafascial, type II or type III radical hysterectomy; radical trachelectomy; with or without pelvic lymphadenectomy; regardless of surgical route via laparotomy or laparoscopy; complete clinical and pathological data. Patients who received non-surgical treatment or neoadjuvant treatment or those with incomplete data were excluded. A total of 12 681 eligible cases were included in the analysis. Figure 1 shows a flow chart of inclusion and the reasons for exclusion.
All patients in our study were clinically staged according to the 2009 FIGO staging system.14 The accuracy of clinical staging was determined in the following steps. First, surgicopathologic parameters were abstracted from the database, then the pathologic stage was assigned based on these parameters with reference to the same 2009 FIGO staging system. Second, case-for-case comparison between clinical stage and pathologic stage was performed using the latter as the reference standard. If they were in accordance, we considered it as accurate; if the surgicopathologic stage was higher than the clinical stage, we considered it clinically understaged regardless of the extent of understaging; and if the surgicopathologic stage was lower than the clinical stage, we considered it clinically overstaged regardless of the extent of overstaging.
The causes of stage inaccuracy were classified as follows: extent of cervical stromal invasion, maximal tumor diameter, vaginal involvement, parametrial infiltration, bladder or rectum invasion, and distant metastases. The extent of cervical stromal invasion refers to tumors confined to the cervix in both clinical staging and pathologic staging; however, the extent of invasion was discrepant. For example, a case was diagnosed pre-operatively as stage IA but a more advanced lesion (infiltration depth >5 mm) was found at post-operative pathologic examination; or a case was pre-operatively diagnosed as stage IB1 but the post-operative pathologic examination showed only micro-invasion or no lesion at all. For cases with more than two causes of stage inaccuracy, the cause with the higher staging implication was attributed. The sequence of staging implication of the causes was as follows: distant metastases > bladder or rectum infiltration > parametrial infiltration > vaginal invasion > maximal tumor diameter > extent of cervical stromal invasion.
All statistical analyses were performed with SPSS Version 17.0 (SPSS, Chicago, Illinois, USA). Continuous variables were expressed as mean (SD) and categorical data as counts and proportions. A χ2 test was used for comparison between groups.
Results
The cases of cervical cancer in our participating hospitals showed an increasing trend, with 2838 cases in 2010–2011 (22.4%), 4114 cases in 2012–2013 (32.4%), and 5729 cases in 2014–2015 (45.2%). The patient characteristics are presented in Table 1. The mean (SD) age of the patients was 47.9 (9.5) years. Overall, squamous cell carcinoma was identified in 11 340 patients (89.4%), adenocarcinoma in 1021 patients (8.05%), adenosquamous carcinoma in 125 patients (1.0%), and other types in 195 patients (1.5%). The distribution of clinical stage was stage IB1 in 7146 patients (56.4%), IIA1 in 2557 patients (20.2%), IA in 1498 patients (11.8%), IB2 in 1041 patients (8.2%), and IIA2 in 439 patients (3.5%).
As shown in Table 2, 1348 patients in stage IA, 6250 patients in stage IB1, 598 patients in stage IB2, 518 patients in stage IIA1, and 112 patients in stage IIA2 were accurately staged. The accuracy of clinical staging was 90.0% in stage IA, 87.5% in stage IB1, 57.4% in stage IB2, 20.3% in stage IIA1, and 25.5% in stage IIA2. The overall accuracy of clinical staging was 69.6%. More clinical overstaging (2666, 21.0%) was observed than clinical understaging (1189, 9.4%). The accuracy of clinical staging decreased significantly as the stage advanced (Table 3).
Analysis of the causes of stage discrepancy revealed that vaginal invasion (2401, 62.3%) was the primary cause, followed by maximal tumor diameter (950, 24.6%), extent of cervical stromal invasion (272, 7.1%), parametrial invasion (223, 5.8%), bladder or rectal infiltration (5, 0.1%), and distant metastases (4, 0.1%). In the sub-analysis of the causes of clinical overstaging, vaginal invasion (2149, 80.6%) was also the primary cause while, in the sub-analysis of clinical understaging, maximal tumor diameter represented the primary cause (565, 47.5%), followed by vaginal invasion (252, 21.2%), parametrial infiltration (223, 18.8%), extent of cervical stromal invasion (140, 11.8%), and advanced lesions (9, 0.7%) (Table 4).
Discussion
Our study aimed to investigate the accuracy of staging early cervical cancer in China. In the analysis of 12 681 patients, the overall accuracy of clinical staging was 69.6%, with 21.0% of the patients being clinically overstaged and 9.4% clinically understaged. The accuracy of clinical staging decreased as clinical stage increased. The accuracy of stage IIA1 and IIA2 were only 20.3% and 25.5%, respectively. As one of the most important aspects of cancer treatment, clinical staging has been studied extensively. Early in 1971 Van Nagell et al stated that discrepancy between clinical staging and pathologic findings is inevitable.15 A significant body of literature has confirmed the efforts in improving the accuracy of clinical staging.16–18 Advanced imaging tests such as MRI, CT or PET have been proposed repeatedly in the staging of cervical cancer and their efficacy has been demonstrated in a number of studies.16 17 19 However, such studies have been excluded from staging classifications until the recent revision. In the FIGO annual report in 2006, the accuracy of stages IA1, IA2, IB1, IB2, and IIA were reported to be 95.0%, 82%, 82%, 61%, and 60%, respectively.20 In another retrospective study of 818 patients with stage IB–IIB, the accuracy of stages IB1, IB2, IIA, and IIB were 85.4%, 77.4%, 35.3%, and 20.5%, respectively.21 Our results are in accordance with these studies, except for the low accuracy of stage IIA.
The accuracy of clinical staging relies on accurate determination of each of the staging parameters. In our study, the causes of inaccurate staging sorted by frequency from most to least were vaginal invasion, maximal tumor diameter, extent of cervical stromal invasion, parametrial invasion, bladder or rectal infiltration, and distant metastases. Because of their unique anatomic position, the cervix and vagina may be evaluated by visual inspection. It has been thought that vaginal invasion can be easily diagnosed by pelvic examination.22 Contrary to this common belief, our study demonstrated that vaginal invasion ranks first as the causes of stage inaccuracy. In stage IIA1 and IIA2, 72.4% and 70.9% of patients were overstaged, respectively.
There are several possible explanations. First, the diagnosis of vaginal invasion may not be as easy as has been thought. Complex clinical scenarios such as the shallowness or disappearance of the vaginal fornix in post-menopausal women, the bulky tumor extending to the vaginal wall, or the inflammation surrounding the tumor make it difficult to determine the degree of vaginal invasion. Canaz et al stated that vaginal invasion can be overestimated by clinical staging in comparison with the pathologic examination.23 Qin et al reported the accuracy of stage IIA as only 35.3% and that more than 50% of the patients with stage IIA were clinically overstaged.21 Second, in previous practice in China the presence of vaginal invasion did not change the treatment modality. Consequently, vaginal invasion may be under-appreciated and the specimen less scrutinized in pathologic examination with false negatives as the consequence.
The efficacy of modern imaging techniques in determining vaginal invasion has only been addressed in a few studies.24–26 The use of vaginal gel in MRI scan has been reported to help the definition of the vaginal fornix on MRI images.18 However, most of these studies are single institutional studies with a limited sample size. More research on the improvement and standardization of these imaging techniques is needed. In addition, biopsy of suspicious vaginal lesions may also be applied to increase the accuracy of staging.
Before the recent FIGO revision, staging of cervical cancer relied solely on pelvic examination in determining tumor diameter. However, pelvic examination alone is limited, especially in patients with an endophytic growth pattern. Compared with pelvic examination, modern imaging tests such as MRI, with its high soft tissue resolution, could delineate the contour of the tumor clearly and overcome the limitation of pelvic examination.17 27 The efficacy of MRI in estimating tumor size has been proved by many previous studies. Therefore, the accuracy of staging maximal tumor diameter can be expected to improve significantly after the revisions.
Parametrial infiltration has been identified as the primary cause of stage inaccuracy in many previous studies.15 21 However, in our study, patients with suspected parametrial infiltration in pelvic examination were staged as IIB or higher and excluded from the analysis. This selection bias explains why parametrial infiltration only ranks fourth in the causes of stage inaccuracy. Similar to parametrial infiltration, bladder or rectum infiltration and distant metastases were also unlikely to be the main cause of stage inaccuracy.
The diagnostic efficacy of modern imaging techniques in determining parametrial infiltration is promising. In a meta-analysis by Thomeer et al, MRI showed a sensitivity of 84% and a specificity of 92% in studies published up to 2011.28 A more recent meta-analysis of the literature between 2012 and 2016 by Woo et al showed that the sensitivity of MRI for determining parametrial infiltration was 0.765 (95% CI 0.67 to 0.84) and specificity was 0.94 (95% CI 0.91 to 0.95).29 Therefore, the accuracy of clinical staging in determining parametrial infiltration may also be improved after the revision of the FIGO staging guidelines.
Extension of cervical stromal invasion ranks third in the causes of stage inaccuracy. One possible explanation for overstaging is that the lesion has been removed by the biopsy. As for understaging, 10% of patients with stage IA were understaged in our study. The high rates of understaging imply that stage IA designation in many patients was based on biopsy instead of cervical conization. However, cervical conization is not mandatory in the diagnosis of stage IA according to the 2009 FIGO guidelines.4 Because modern imaging tests—including MRI with its high soft tissue resolution—are not efficient in diagnosing microinvasion,30 no significant impact can be expected after the revision of the staging system. The high rate of understaging in stage IA suggests that cervical conization should be routinely performed in this stage.
We recognize that this study has important limitations, including selection bias. Patients with neoadjuvant chemotherapy were excluded from the analysis. Also, tumor specimens did not undergo central pathologic review. Therefore, some degree of tumor stage misclassification may have occurred. Although previous studies classify pelvic lymph node metastasis as stage IIIB, it was not analyzed in our study because pelvic lymph node metastasis could not be evaluated by pelvic examination or other FIGO staging procedures before the recent revision.
Despite these limitations, our results reflect the performance of clinical staging in China before the recent revision of the FIGO staging system. The strengths of the study include the large number of patients and the fact that it was a multicenter effort. The hospitals participating in this study were all tertiary hospitals. They provide care to local women with cervical cancer and are the mainstay of cancer treatment in China, so the accuracy of clinical staging in these hospitals can reflect the standards of practice in China.
While questions remain as to the performance of the revised staging system, our study provides a perspective of clinical staging in early cervical cancer in China before the FIGO staging revision. As accuracy was the most important indicator of staging quality, the quality of staging in early cervical cancer in China can be regarded as sub-optimal. Enhancing quality control in staging and further optimization of the staging system are both essential to improve the accuracy of clinical staging.
References
Footnotes
Contributors CLC contributed to the design of the study, data collection, data interpretation, and revising the manuscript. WFZ contributed to the design of the study, data collection and analysis, and writing the manuscript. PL contributed to data interpretation and revising the manuscript, WLL, MH, WDZ, AWL, and YN contributed to data collection, interpretation and revising the manuscript. All authors approved the final manuscript and agreed to be accountable for all aspects of the work.
Funding This work was supported by grants from the National Science & Technology Pillar Program of China during the 12th Five-year Plan period (2014BAI05B03), Key project of the Science Program of Guangzhou for Collaborative Health and Medicine Innovation (201508020264), and the Natural Science Foundation of Guangdong, China (2015A030311024).
Competing interests None declared.
Patient consent for publication Not required.
Provenance and peer review Not commissioned; externally peer reviewed.
Data availability statement Data are available upon reasonable request. All data relevant to the study are included in the article or uploaded as supplementary information.