Prediction of 10-year Overall Survival in Patients with Operable Cervical Cancer using a Probabilistic Neural Network

A Case for the Conservative Management of Stage IA Cervical Cancer

Cervical cancer remains a significant public health concern within the United States and across the world. Cervical cancer is most frequently diagnosed in women between the ages of 35 and 44 and therefore affects a younger patient population than many other cancers. The management of early-stage disease has frequently utilized radical hysterectomy with the associated increased surgical morbidity, without clear evidence of any benefits. In stage IA disease, there are retrospective pathologic data supporting the safety of conservative surgery and lymphadenectomy over radical hysterectomy. There are also emerging prospective studies supporting conservative management. This editorial presents the evidence for conservative management of stage IA cervical cancer by reviewing the existing retrospective studies as well as the ongoing prospective studies.

Systematic review and meta-analysis of prediction models used in cervical cancer

BACKGROUND

Cervical cancer is one of the most common cancers in women with an incidence of around 6.5 % of all the cancer in women worldwide. Early detection and adequate treatment according to staging improve the patient's life expectancy. Outcome prediction models might aid treatment decisions, but a systematic review on prediction models for cervical cancer patients is not available.

DESIGN

We performed a systematic review for prediction models in cervical cancer following PRISMA guidelines. Key features that were used for model training and validation, the endpoints were extracted from the article and data were analyzed. Selected articles were grouped based on prediction endpoints i.e. Group1: Overall survival, Group2: progression-free survival; Group3: recurrence or distant metastasis; Group4: treatment response; Group5: toxicity or quality of life. We developed a scoring system to evaluate the manuscript. As per our criteria, studies were divided into four groups based on scores obtained in our scoring system, the Most significant study (Score > 60 %); Significant study (60 % > Score > 50 %); Moderately Significant study (50 % > Score > 40 %); least significant study (score < 40 %). A meta-analysis was performed for all the groups separately.

RESULTS

The first line of search selected 1358 articles and finally 39 articles were selected as eligible for inclusion in the review. As per our assessment criteria, 16, 13 and 10 studies were found to be the most significant, significant and moderately significant respectively. The intra-group pooled correlation coefficient for Group1, Group2, Group3, Group4, and Group5 were 0.76 [0.72, 0.79], 0.80 [0.73, 0.86], 0.87 [0.83, 0.90], 0.85 [0.77, 0.90], 0.88 [0.85, 0.90] respectively. All the models were found to be good (prediction accuracy [c-index/AUC/R²] >0.7) in endpoint prediction.

CONCLUSIONS

Prediction models of cervical cancer toxicity, local or distant recurrence and survival prediction show promising results with reasonable prediction accuracy [c-index/AUC/R² > 0.7]. These models should also be validated on external data and evaluated in prospective clinical studies.

Methodological conduct of prognostic prediction models developed using machine learning in oncology: a systematic review

BACKGROUND

Describe and evaluate the methodological conduct of prognostic prediction models developed using machine learning methods in oncology.

METHODS

We conducted a systematic review in MEDLINE and Embase between 01/01/2019 and 05/09/2019, for studies developing a prognostic prediction model using machine learning methods in oncology. We used the Transparent Reporting of a multivariable prediction model for Individual Prognosis Or Diagnosis (TRIPOD) statement, Prediction model Risk Of Bias ASsessment Tool (PROBAST) and CHecklist for critical Appraisal and data extraction for systematic Reviews of prediction Modelling Studies (CHARMS) to assess the methodological conduct of included publications. Results were summarised by modelling type: regression-, non-regression-based and ensemble machine learning models.

RESULTS

Sixty-two publications met inclusion criteria developing 152 models across all publications. Forty-two models were regression-based, 71 were non-regression-based and 39 were ensemble models. A median of 647 individuals (IQR: 203 to 4059) and 195 events (IQR: 38 to 1269) were used for model development, and 553 individuals (IQR: 69 to 3069) and 50 events (IQR: 17.5 to 326.5) for model validation. A higher number of events per predictor was used for developing regression-based models (median: 8, IQR: 7.1 to 23.5), compared to alternative machine learning (median: 3.4, IQR: 1.1 to 19.1) and ensemble models (median: 1.7, IQR: 1.1 to 6). Sample size was rarely justified (n = 5/62; 8%). Some or all continuous predictors were categorised before modelling in 24 studies (39%). 46% (n = 24/62) of models reporting predictor selection before modelling used univariable analyses, and common method across all modelling types. Ten out of 24 models for time-to-event outcomes accounted for censoring (42%). A split sample approach was the most popular method for internal validation (n = 25/62, 40%). Calibration was reported in 11 studies. Less than half of models were reported or made available.

CONCLUSIONS

The methodological conduct of machine learning based clinical prediction models is poor. Guidance is urgently needed, with increased awareness and education of minimum prediction modelling standards. Particular focus is needed on sample size estimation, development and validation analysis methods, and ensuring the model is available for independent validation, to improve quality of machine learning based clinical prediction models.

The annual recurrence risk model for tailored surveillance strategy in patients with cervical cancer

PURPOSE

Current guidelines for surveillance strategy in cervical cancer are rigid, recommending the same strategy for all survivors. The aim of this study was to develop a robust model allowing for individualised surveillance based on a patient's risk profile.

METHODS

Data of 4343 early-stage patients with cervical cancer treated between 2007 and 2016 were obtained from the international SCCAN (Surveillance in Cervical Cancer) consortium. The Cox proportional hazards model predicting disease-free survival (DFS) was developed and internally validated. The risk score, derived from regression coefficients of the model, stratified the cohort into significantly distinctive risk groups. On its basis, the annual recurrence risk model (ARRM) was calculated.

RESULTS

Five variables were included in the prognostic model: maximal pathologic tumour diameter; tumour histotype; grade; number of positive pelvic lymph nodes; and lymphovascular space invasion. Five risk groups significantly differing in prognosis were identified with a five-year DFS of 97.5%, 94.7%, 85.2% and 63.3% in increasing risk groups, whereas a two-year DFS in the highest risk group equalled 15.4%. Based on the ARRM, the annual recurrence risk in the lowest risk group was below 1% since the beginning of follow-up and declined below 1% at years three, four and >5 in the medium-risk groups. In the whole cohort, 26% of recurrences appeared at the first year of the follow-up, 48% by year two and 78% by year five.

CONCLUSION

The ARRM represents a potent tool for tailoring the surveillance strategy in early-stage patients with cervical cancer based on the patient's risk status and respective annual recurrence risk. It can easily be used in routine clinical settings internationally.

Predicting Cervical Cancer Outcomes: Statistics, Images, and Machine Learning

Cervical cancer is a very common and severe disease in women worldwide. Accurate prediction of its clinical outcomes will help adjust or optimize the treatment of cervical cancer and benefit the patients. Statistical models, various types of medical images, and machine learning have been used for outcome prediction and obtained promising results. Compared to conventional statistical models, machine learning has demonstrated advantages in dealing with the complexity in large-scale data and discovering prognostic factors. It has great potential in clinical application and improving cervical cancer management. However, the limitations of prediction studies and prediction models including simplification, insufficient data, overfitting and lack of interpretability, indicate that more work is needed to make clinical outcome prediction more accurate, more reliable, and more practical for clinical use.

Novel artificial intelligence machine learning approaches to precisely predict survival and site-specific recurrence in cervical cancer: A multi-institutional study

BACKGROUND

Machine learning (ML) has been gradually integrated into oncologic research but seldom applied to predict cervical cancer (CC), and no model has been reported to predict survival and site-specific recurrence simultaneously. Thus, we aimed to develop ML models to predict survival and site-specific recurrence in CC and to guide individual surveillance.

METHODS

We retrospectively collected data on CC patients from 2006 to 2017 in four hospitals. The survival or recurrence predictive value of the variables was analyzed using multivariate Cox, principal component, and K-means clustering analyses. The predictive performances of eight ML models were compared with logistic or Cox models. A novel web-based predictive calculator was developed based on the ML algorithms.

RESULTS

This study included 5112 women for analysis (268 deaths, 343 recurrences): (1) For site-specific recurrence, larger tumor size was associated with local recurrence, while positive lymph nodes were associated with distant recurrence. (2) The ML models exhibited better prognostic predictive performance than traditional models. (3) The ML models were superior to traditional models when multiple variables were used. (4) A novel predictive web-based calculator was developed and externally validated to predict survival and site-specific recurrence.

CONCLUSION

ML models might be a better analytic approach in CC prognostic prediction than traditional models as they can predict survival and site-specific recurrence simultaneously, especially when using multiple variables. Moreover, our novel web-based calculator may provide clinicians with useful information and help them make individual postoperative follow-up plans and further treatment strategies.