ABSTRACT
The prediction of prognosis using nomogram in cervix cancer
Seo, Young Seok
Dept. of Radiation oncology
The Graduate School
Hanyang University
Purpose
It could be assumed that the present criteria of the International Federation of Gynecology and Obstetrics (FIGO) staging are insufficient to predict patients’ survival because it is based on only limited conventional procedures. Recently, nomograms have been introduced as a tool for estimating prognosis. The purpose of the present study is to develop a nomogram to predict the probability of a 5-year survival after radical radiotherapy (RT) for uterine cervix cancer.
Methods
We retrospectively analyzed 668 patients who underwent radical RT for uterine cervix cancer between March 1994 and April 2002 in Korea Institute of Radiological & Medical Sciences. Multivariate analysis using Cox proportional hazards regression was done and this Cox model was used as the basis for the nomogram. Clinicopathological variables included age, hemoglobin level before RT, FIGO stage, maximal diameter of tumor, lymph node status, concurrent chemoradiotherapy, and dose of RT on point-A. The model was internally validated for discrimination and calibration using bootstrap resampling.
Results
On multivariate regression analysis, the model showed that age, hemoglobin level before RT, FIGO stage, maximal diameter of tumor, lymph node status, concurrent chemoradiotherapy, and dose of RT on point-A were significant predictor for overall survival. The survival prediction model demonstrated good calibration and discrimination. The bootstrap-corrected concordance index was 0.73. When compared with the predictive ability of FIGO stage, the nomogram discrimination was superior (P=0.01)
Conclusion
The nomogram represented a significantly better discrimination than FIGO staging system. This improved prediction of survival probability would be useful in counseling patients, determining treatment modality and schedule, and clinical trial design. Before this nomogram become universally acceptable, it needs external validation which is tested on other hospitals’ database using this nomogram.