![The World Congress on Controversies in Obstetrics, Gynecology and Infertility (COGI). Click here to learn more.]({"type":"image" , "src" : "/sda/112382/JournalAd-COGI2024-Leaderboard.jpg"})
Abstract
Up to now, there are no relevant studies on prognostic factors of cervical mucinous adenocarcinoma. Therefore, we explored the prognostic factors for cervical mucinous adenocarcinoma, and established and validated the prognostic model using the SEER database. We selected the independent factors through univariate and multivariate analyses. LASSO regression analysis was conducted to identify potential risk factors. In conjunction with LASSO and multivariate analysis, the nomogram incorporated three variables, including age, tumour size, and AJCC stage for OS. The c-index was 0.794 and 0.831 in development and validated cohorts, indicating that this prediction model showed adequate discriminative ability in the development cohort. Besides, calibration curves showed good concordance for the development cohort, as well as the validation cohort. We constructed a first-of-its-kind nomogram to predict cervical mucinous adenocarcinomas OS and it showed better performance than AJCC and FIGO stages. Patients with cervical mucinous adenocarcinoma might benefit from using this model to develop tailored treatments.
What is already known on this subject? Cervical cancer has a variety of pathological types. The biological behaviour of each type is different, and the prognosis is quite different.
What do the results of this study add? We analysed and explored the relevant factors affecting the prognosis of cervical mucinous adenocarcinoma.
What are the implications of these findings for clinical practice and/or further research? Through the analysis of the SEER dataset, the prognostic factors affecting cervical mucinous adenocarcinoma were identified, and the first predictive model was created to predict the prognosis to help doctors develop individualised treatment plans and follow-up plans.
IMPACT STATEMENT
Introduction
In the world, cervical cancer was still a problem endangering women’s lives (Siegel et al. Citation2021, Sung et al. Citation2021). Adenocarcinoma is the second most common histological type of cervical cancer, and increasing in real and relative incidence in the past 20 years (Sung et al. Citation2021). Mucinous adenocarcinoma is the third most common cervical adenocarcinoma, after the usual type and gastric type (Stolnicu et al. Citation2018).
Currently, the treatment and prognostic factors for cervical adenocarcinoma are the same as those for squamous cell carcinomas. In recent years, increasing evidence suggests that AC behaves differently to squamous cell carcinomas, with different responses to treatment and different prognostic factors. Hu et al. reported that cervical adenocarcinoma patients have significantly lower overall survival (OS) than those with squamous cell carcinomas (Hu et al. Citation2018). Up to now, there are no relevant studies on prognostic factors of cervical mucinous adenocarcinoma. Thus, a predictive model for cervical mucinous adenocarcinoma should be developed to predict prognosis and optimise treatment strategies. However, the low incidence of cervical mucinous adenocarcinoma prevented studies on its prognosis and treatment outcomes from being conducted. Based on the Surveillance, Epidemiology, and End Results (SEER) database, we explored factors that predict OS for cervical mucinous adenocarcinoma, and an individualised prognostic model for optimal clinical management was built and validated.
In recent years, multiple prognostic factors have been incorporated into nomograms to accurately predict survival rates. Various cancers such as cervical cancer with lymph node metastasis have been predicted using it (Yi et al. Citation2022). Nomograms are excellent prognostic tools, as they can precisely predict the patient final survival rate based on clinical variables. The patient’s 1-, 3-, and 5-year OS for cervical mucinous adenocarcinoma has been estimated using nomograms, which could be beneficial in personalised treatments and improved outcomes.
Methods
Patient population
SEER*Stat 8.3.9 software was used to extract data from the SEER Research Plus Database, 18 Registries, Nov 2020 Sub (2000–2018). Since SEER is a public database, this study is exempt from ethical review and does not require patients to sign informed consent. Histology codes were:8480/3, 8481/3, and 8482/3, according to ICD-O-3 (Fritz et al. Citation2000, Kurman et al. Citation2014).
Inclusion criteria were: histologically proven malignant mucinous adenocarcinoma between 2000 and 2018; known survival months, tumour size, and the number of lymph nodes examined. As criteria for exclusion, squamous cell carcinomas, other adenocarcinomas, unknown tumour sizes, and no survival time were used. The missing data were handled with a complete-case analysis. As a public database, the SEER database did not require institutional review board approval. Access to the dataset and permission to use it was granted by the SEER program.
We used 320 patients diagnosed between 2000 and 2012 to build the model, and 256 patients diagnosed between 2013 and 2018 to validate the model. Figure S1 illustrates the process flow. SEER data were collected and analysed as follows: age, race, diagnosed year, histologic, grade, American Joint Committee on Cancer (AJCC) stage system tumour, node, metastasis (TNM) stage, surgical options, radiation, chemotherapy, tumour size, the number of lymph nodes examined (LNE), the number of positive lymph nodes (LNP), sequence number of multi-primary tumours, survival time, and survival status. Age was classified as: <40, 40–60, ≥60 years. The tumour size was divided into three categories: ≤2, 2–4, and >4 cm (Bhatla et al. Citation2018). LNE was classified as 0, 1–15, and >15, and LNP was grouped as 0, 1–3, and >3, based on previous studies (Kwon et al. Citation2018, Ni et al. Citation2021). A patient’s OS time is defined as the time from diagnosis until death or the date of deletion of data. OS was the study’s endpoint. A Declaration of Helsinki was followed in this study (revised in 2013) (WMA Citation2013). And we confirmed the Enhancing the QUAlity and Transparency Of health Research (EQUATOR) network guidelines.
Figure 1. Forest plot demonstrating the multivariate Cox regression model for predicting OS in the development cohort. OS, overall survival. AJCC stage: American Joint Committee on Cancer (AJCC) staging system; Size: tumour size; Age: age of diagnosis.
Model constructed
LASSO analysis was conducted to identify potential risk factors preventing variables from being overfitted and penalising regression coefficients’ absolute values in the development cohort. Large penalties caused estimates of weaker factors to shrink towards zero, resulting in only the most effective factors remaining in the model. The development cohort was analysed via LASSO regression to determine the most significant predictors. For the stability of the model, we adjusted the penalty coefficient to 1 standard error. Finally, eight variables were selected for OS (Figure S2(A,B)). The intersection of variables of independent risk factors and Lasso regression were included in the final model. In the end, three variables were used to construct the model. There was no difference in the selected variables between the development and validation cohorts.
Figure 2. (A) Nomogram for predicting 1-, 3- and 5-year OS in cervical mucinous adenocarcinoma patients. (B) An example of how to use nomogram for cervical mucinous adenocarcinoma patients. futime: survival time(months).
Statistical analysis
Since continuous variables are not symmetrical in our study, we use the median and interquartile range (IQR) values for description. In categorical variables, the number of each group and its proportion is described. Our independent factors were selected via univariate and multivariate regression analyses. Nomograms were constructed based on the Cox proportional hazard results to visualise and intuitively see the proportion of each variable on the estimated 1-, 3- and 5-year OS probability. The nomogram was experienced internally and in temporal validation. Using a prediction model, we also calculated the risk score. Then, we divided patients into a low- or high-risk group according to median score. We calculated survival curves using Kaplan-Meier methods with log-rank tests to identify significant differences. In addition, we evaluated the nomogram predictive performance by receiver operating characteristic (ROC) curves, c-index, and calibration curves. All analysis used R software version 4.1.3 (http://www.R-project.org) in our study. p value <0.05 was considered statistically significant.
Results
Clinical characteristics
We involved 576 cervical mucinous adenocarcinoma cases diagnosed between 2000 and 2018 from the SEER database. This model was developed using data from patients diagnosed between 2000 and 2012 (n = 320), with data from patients diagnosed from 2013 to 2018 (n = 256) used for temporal validation. summarised patients baseline characteristics.
Table 1. Characteristics of the development cohort and validation cohort.
Identify independent prognostic factors of OS in development cohort
Cox regression analyses, both univariate and multivariate, were conducted to determine independent factors. According to the univariate analysis, age, race, histological type, grade, AJCC stage, T, N, M stage, FIGO stage, tumour size, surgery, chemotherapy, radiotherapy, LNE, and LNP were linked to OS. Meanwhile, based on multivariate analysis, age, race, chemotherapy, radiation, tumour size (cm), AJCC stage, FIGO stage, and M stage were independent prognostic factors ().
Table 2. Univariable and multivariable analysis for cervical mucinous adenocarcinoma patients.
Develop prognostic nomogram for OS
LASSO regression analysis identified eight variables including age, tumour size, LNE, LNP, grade, surgical options, T stage, and AJCC stage with 1 standard error criterion coefficient value. With the results of LASSO and multivariate analysis, three variables, age, tumour size, and AJCC stage for OS, were included in the construction of the static nomogram.
The forest plot depicted the multivariate Cox regression model for predicting OS in development cohorts (). In comparison with AJCC stage I, stage II, III, and IV were associated with poor prognosis (HR = 2.26, 95% CI: 1.31–3.92, p = 0.004; HR = 3.87, 95% CI: 2.36–6.33 p < 0.001; HR = 6.68, 95% CI: 4.01–11.11, p < 0.001, respectively). With regard to tumour size, tumour size ≤2 cm as a reference, tumour size arranged 2–4 cm and tumour size >4 cm were linked to a relatively worse prognosis (HR = 1.55, 95% CI: 0.90–2.26, p = 0.112; HR = 3.12 95% CI:1.88–5.16, p < 0.001, respectively). In comparison with age <40 years, 40–60years and ≥60 years were associated with poor prognosis (HR = 1.14, 95% CI: 0.89–2.22, p = 0.144; HR = 2.12, 95% CI: 1.26–3.58 p = 0.005, respectively).
Variables’ contributions to survival outcomes are represented by the length of the lines in the nomogram. In the ‘Points’ scale, every subtype of variable makes up a point in the nomogram. Adding each subtype’s score results in the patient’s total score. From these total scores on the ‘Total points’ scale, a straight line was drawn, giving each patient 1-, 3-, and 5-year OS probabilities (). Besides, showed each prognostic factor’s score of the nomogram. And the AJCC stage scores range from 0 to 100 points, tumour size from 0 to 60 points, and age from 0 to 40 points.
Table 3. Each prognostic factor’s score of OS nomogram.
Using the nomogram, clinicians could predict the survival rates of individual patients. For instance, the patient, 50 years old (18 points) with a primary tumour measured 1 cm (0 points), AJCC stage I (0 points), had 18 points in total. OS less than 1-, 3- and 5-year probabilities were 1.91%,5.78%, and 9.39% respectively ().
Internal and temporal validation of the nomogram
In different cohorts, the nomogram was assessed by the ROC curves. In the range of 0.5–1, the greater the value of the area under the curve (AUC) was, the more accurate and robust the nomogram was. In development cohort, AUC values for 1-, 3-, and 5-year OS probabilities were respectively 0.856, 0.854, and 0.847 () and in validation cohort AUC values were 0.862,0.846 and 0.835, respectively (). The values both in the development and validation cohorts were larger than AJCC and FIGO stage ().
Figure 3. Nomogram, AJCC stage and FIGO stage ROC curves for 1-, 3- and 5-year OS (A, C, E) in development cohort and (B, D, F) validation cohort, respectively. ROC, receiver operator characteristic.
C-index ranging from 0.5 to 1, the greater the value of the c-index was, the more accurate prediction of the model was. Based on the c-index of 0.794 in development cohort, this prediction model showed adequate discriminative ability. In the temporal validation, the c-index for the nomogram was 0.831. The c-index for OS predicted by the AJCC stage and FIGO stage was 0.753 and 0.685 in the development cohort. In the validation cohort, the c-index for OS predicted by the AJCC stage and FIGO stage was 0.772 and 0.745. This showed that our model is superior to AJCC stage and FIGO stage.
Based on the calibration plot, the predicted probability of 1-, 3-, and 5-year OS was compared with the observed probability. Results showed good concordance between the 1-, 3-, and 5-year OS nomograms for the development cohort, as well as the validation cohort (Figure S3(A–F)). Almost perfect calibration curves were observed.
Using prognostic scores, participants in the development cohort were divided into low-risk and high-risk groups to assess the predictability and validity of the prediction model (Figure S4(A)). Using the same risk cut-off values in the validation cohort, Kaplan-Meier survival curves were generated, indicating significantly different prognoses (p < 0.001), illustrating that patients with high mortality risks were identified by the model (Figure S4(B)).
Discussion
In recent years, the risk factors affecting the prognosis of cervical cancer have been widely studied, however, most studies did not carefully distinguish the pathological types, and only analysed the pathological types as a variable. Different pathological types of cervical cancer have different high-risk factors affecting prognosis due to their different biological behaviours.
At the same time, cervical mucinous adenocarcinoma has the characteristics of strong invasiveness and difficulty in early diagnosis, so the common high-risk factors based on other types of cervical cancer are not necessarily applicable to cervical mucinous adenocarcinoma. In this study, we focussed on the specific pathological type, cervical mucinous adenocarcinoma, and used the SEER dataset to further analyse the high-risk factors affecting the prognosis of cervical mucinous adenocarcinoma patients to accurately predict the prognosis of patients and develop individualised anti-tumour treatment and follow-up plans.
In this study, a variety of clinical factors affecting the prognosis of patients were included. Through univariate and multivariate analysis, the independent risk factors affecting the prognosis of cervical mucinous adenocarcinoma were identified. Combined with the results of LASSO analysis, the model predicted 1-year, 3-year 5-year survival was established. An individual’s survival probability at a given point in time can be estimated by nomogram. The nomogram was a graphics-based prognostic tool that displayed clinical results. Several clinical variables are accounted for in its calculation to ensure the prediction for patients’ OS is accurate. Ni et al. calculated a nomogram based on tumour grade, stage T, N, M, tumour size, and surgery to predict cervical adenocarcinoma cancer-specific survival (Ni et al. Citation2021). Bogani et al. developed a nomogram to predict the risk of recurrence of cervical dysplasia, and margin status and HPV persistence were important factors (Bogani et al. Citation2022). Therefore, we aimed to create a nomogram to predict the prognosis for patients with cervical mucinous adenocarcinoma.
Multivariate analysis indicated that age, race, chemotherapy, radiation, tumour size, AJCC stage, FIGO stage, and M stage were independent prognostic factors for cervical mucinous adenocarcinoma patients’ OS. According to other studies in all cervical cancer pathological types, age and race affect the prognosis (C. Wang et al. Citation2018, Zhou et al. Citation2018). Xie et al. found that older patients have lower 1- and 5-year cancer-specific survival rates than younger patients (Xie et al. Citation2020). It has been widely acknowledged that race also affects tumour prognosis due to genetic differences (Liu et al. Citation2020, Z. Wang et al. Citation2021). The late stage often means a poor prognosis and we found distant metastasis and late staging are independent risk factors affecting prognosis. So early diagnosis of cervical cancer is very important for a patient’s prognosis, and several cancer markers have been developed to detect invasive forms, including CEA, SCC-Ag, and CD44 (Valenti et al. Citation2017). Bizzarri et al. also investigated the prevalence of peritoneal human papillomavirus (PHP) infection in different clinical cervical cancer settings and found that a higher frequency of PHP was documented in patients with peritoneal carcinomatosis (Bizzarri et al. Citation2021). These markers may also be useful in other fields other than early diagnoses, such as evaluation and monitoring of treatments to improve diagnosis and treatment of cervical cancer. Another independent prognostic factor was tumour size. Treatment strategies were incorporated into the univariate analysis for our study. As depicited in , cases with surgery improved outcomes. It was concluded that a high tumour burden (tumour size) would mean a poor prognosis, whereas surgery led to improved outcomes. For surgical treatment of early-stage cervical cancer (ECC), minimally invasive surgery (MIS) is debated against open surgery. The Laparoscopic Approach to Cervical Cancer (LACC) trial reported inferior oncological outcomes and lower survival rates for women who underwent MIS compared to open surgery (Ramirez et al. Citation2018). However, a recent study showed no significant difference between MIS and open surgery for women with FIGO stage Ib1-IIa2 cervical adenocarcinoma (Giannini et al. Citation2022, Zhu et al. Citation2022). Nevertheless, for a better understanding of the MIS effect on ECC survival rates, larger series needs to be analysed. LNP was involved in poor prognosis, and LNE was associated with improved outcomes but was not an independent prognostic factor for the OS of our study. Zhou et al. analysed 312 cervical adenocarcinomas found that positive pelvic nodes and age at surgery were independent prognostic factors (Zhou et al. Citation2018). The FIGO staging system is widely used in cervical cancer. However, this study found that the AJCC staging system has a greater advantage in evaluating the prognosis than the FIGO staging system in cervical mucinous adenocarcinoma. Therefore, it is more important to perform AJCC staging for patients with cervical mucinous adenocarcinoma.
To further evaluate the prognosis of patients, it is necessary to comprehensively consider various factors affecting the prognosis. Clinically, FIGO staging, TNM staging, AJCC staging, and other staging systems are widely used to evaluate the prognosis of malignant tumours. However, previous studies have found that clinical variables affecting patient prognosis are not limited to the tumour stage. Epidemiological factors such as age, race, operation, radiotherapy, chemotherapy and other interventions are closely related to the prognosis of patients. Therefore, the staging system cannot be used for individualised prediction. Taking all independent factors into account is necessary for accurate prediction. We established prognostic nomograms by combining multivariate with LASSO analysis results for 1-, 3-, and 5-year OS. Consequently, these factors were taken into consideration to develop a nomogram: AJCC stage, tumour size, and age. Clinical practicability and predictive performance were assessed by the ROC curve and c-index. For probabilities of 1-, 3- and 5-year OS, the AUC values were 0.856, 0.8554, 0.8547 in development cohort and 0.862, 0.8446, and 0.8553 respectively in validation cohort. The prediction accuracy for OS was evaluated via a calibration curve, which was well correlated with actual results. Additionally, the c-index of the nomogram was 0.794, 0.831 in the development and validation cohort. As a result, the nomogram appears to be highly predictive. As compared to the AJCC stage and FIGO stage, it proved to be a more accurate method of prediction. It will be possible to design individualised treatment plans for cervical mucinous adenocarcinoma patients using this nomogram.
The model quantified and visualised each prognostic factor to predict the patients’ 1-, 3-, and 5-year survival chances via static nomograms. Using our prediction model, doctors can identify each person’s risk based on a few known factors, predict the outcome, and prescribe treatments and follow-ups for cervical mucinous adenocarcinoma patients. Because high-risk patients are more likely to die, they should receive comprehensive treatment and be closely monitored. Studies also included surgery, chemotherapy, and radiotherapy as treatment modalities for their effect on the prognosis of cervical mucinous adenocarcinoma. It turn out that surgery was beneficial, but not an independent factor.
There were some limitations. Firstly, retrospective analysis of the model may have introduced bias due to a lack of random assignment. Secondly, there were some potential independent risk factors omitted from the SEER database, including parametrial involvement and LVSI. Because of information shortages, it was impossible to fully examine each of the prognostic factors. Thirdly, despite the model’s validation internally and temporally with the SEER database, more institutions need to evaluate it externally. Additionally, further randomised studies are needed, which will be carried out in the future.
In general, a few nomograms have been developed for cervical cancer, but cervical mucinous adenocarcinoma has not been studied specifically due to its rarity. In this study, we developed the first prognostic model for cervical mucinous adenocarcinoma to predict 1-, 3-, and 5-year OS, and it showed better performance than AJCC and FIGO stages. Patients with cervical mucinous adenocarcinoma might benefit from using this model to receive tailored and individualised treatments.
Ethics statement
The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.
Authors contributions
Contributions: (I) Conception and design: B Cui and W Zhang; (II) Administrative support: B Cui and W Zhang; (III) Collection and assembly of data: Q Liu, R Li, Z Mao, N Jiang, B Wang; (IV) Data analysis and interpretation: Y Hao, Q Liu; (V) Manuscript writing: All authors; (VI) Final approval of manuscript: All authors.
Supplemental Material
Download Zip (6.7 MB)Disclosure statement
No potential conflict of interest was reported by the author(s).
Additional information
Funding
References
- Bhatla, N., et al., 2018. Revised FIGO staging for carcinoma of the cervix uteri. International Journal of Gynecology & Obstetrics, 145, 129–135.
- Bizzarri, N., et al., 2021. PeritonealHPV‐DNA test in cervical cancer (PIONEER study): a proof of concept. International Journal of Cancer, 148 (5), 1197–1207.
- Bogani, G., et al., 2022. Development of a Nomogram Predicting the Risk of Persistence/Recurrence of Cervical Dysplasia. Vaccines, 10 (4), 579.
- Fritz, A., et al., 2000. International classification of diseases for oncology. 3rd ed. Geneva: World Health Organization. https://apps.who.int/iris/handle/10665/42344.
- Giannini, A., et al., 2022. Minimally Invasive Surgery for Cervical Cancer: Should We Look beyond Squamous Cell Carcinoma? Journal of Investigative Surgery : The Official Journal of the Academy of Surgical Research, 35 (7), 1602–1603.
- Hu, K., et al., 2018. Comparison of treatment outcomes between squamous cell carcinoma and adenocarcinoma of cervix after definitive radiotherapy or concurrent chemoradiotherapy. Radiation Oncology, 13 (1), 249.
- Kurman, R. J., et al., 2014. WHO classification of tumours of female reproductive organs. 4th ed. Lyon: International Agency for Research on Cancer.
- Kwon, J., et al., 2018. The Prognostic Impact of the Number of Metastatic Lymph Nodes and a New Prognostic Scoring System for Recurrence in Early-Stage Cervical Cancer with High Risk Factors: A Multicenter Cohort Study (KROG 15-04). Cancer Research and Treatment, 50 (3), 964–974.
- Liu, Q., et al., 2020. Development and validation of a SEER-based prognostic nomogram for cervical cancer patients below the age of 45 years. Bosnian Journal of Basic Medical Sciences, 21 (5), 620–631.
- Ni, X., et al., 2021. Development and validation of a novel nomogram to predict cancer-specific survival in patients with uterine cervical adenocarcinoma. Annals of Translational Medicine, 9 (4), 293–293.
- Ramirez, P. T., et al., 2018. Minimally Invasive versus Abdominal Radical Hysterectomy for Cervical Cancer. The New England Journal of Medicine, 379 (20), 1895–1904.
- Siegel, R. L., et al., 2021. Cancer Statistics, 2021. CA: A Cancer Journal for Clinicians, 71 (1), 7–33.
- Stolnicu, S., et al., 2018. International Endocervical Adenocarcinoma Criteria and Classification (IECC). The American Journal of Surgical Pathology, 42 (2), 214–226.
- Sung, H., et al., 2021. Global Cancer Statistics 2020: GLOBOCAN Estimates of Incidence and Mortality Worldwide for 36 Cancers in 185 Countries. CA: a Cancer Journal for Clinicians, 71 (3), 209–249.
- Valenti, G., et al., 2017. Tumor markers of uterine cervical cancer: a new scenario to guide surgical practice? Updates in Surgery, 69 (4), 441–449.
- Wang, C., et al., 2018. A Prognostic Nomogram for Cervical Cancer after Surgery from SEER Database. Journal of Cancer, 9 (21), 3923–3928.
- Wang, Z., et al., 2021. A nomogram prediction model for lymph node metastasis in endometrial cancer patients. BMC Cancer, 21 (1), 748.
- WMA, 2013. World Medical Association Declaration of Helsinki: ethical principles for medical research involving human subjects. JAMA, 310 (20), 2191–2194.
- Xie, S., et al., 2020. Characteristics and treatments of patients aged 65 years or over with cervical cancer. Clinical Interventions in Aging, 15, 841–851.
- Yi, J., et al., 2022. Development and validation of novel nomograms to predict the overall survival and cancer-specific survival of cervical cancer patients with lymph node metastasis. Frontiers in Oncology, 12, 857375.
- Zhou, J., et al., 2018. Postoperative clinicopathological factors affecting cervical adenocarcinoma. Medicine, 97 (2), e9323.
- Zhu, H., et al., 2022. The impact of minimally invasive surgery on treating patients with early cervical adenocarcinoma. Journal of Investigative Surgery, 35 (7), 1593–1601.