AIM: Using parameters obtained from magnetic resonance imaging (MRI), we constructed a prediction model for parametrial invasion (PMI) of cervical cancer and validated the model in different sets of patients. PATIENTS AND METHODS: Retrospectively, 251 patients with cervical cancer stages IA2-IIA, who had received a radical hysterectomy, were assigned to training and validation cohorts. After the development of the scoring index using logistic coefficient analysis, the performance of the prediction model was assessed using independent validation sets. RESULTS: In the training cohort (n = 167), multivariate analysis indicated that the patient's stage, the cephalocaudal tumor diameter measured by MRI, and finding of PMI as obtained by MRI were independent predictors of PMI (P = 0.010, < 0.001, and 0.020, respectively). These predictors were internally validated by a rigorous bootstrapping method with statistical significance. The scoring index was created based on logistic coefficients, and the maximal score yielding a negative likelihood ratio less than 0.05 was selected as a cutoff. The cutoff was translated into the following criteria identifying a very low-risk group for PMI: (1) FIGO stage IA2-IB1, (2) no MRI finding suggesting PMI, and (3) cephalocaudal tumor diameter less than 1.0 cm by MRI. The negative predictive value (NPV) was 98.5% (95% confidence interval [CI]=91.7% to 100%). In the external validation cohort (n = 84), the NPV was 100% (95% CI = 90% to 100%). CONCLUSION: The current prediction model showed reliable performance for the identification of patients at low risk for PMI. It may be useful for stratification of patients and evaluation of results in future trials.
AIM: Using parameters obtained from magnetic resonance imaging (MRI), we constructed a prediction model for parametrial invasion (PMI) of cervical cancer and validated the model in different sets of patients. PATIENTS AND METHODS: Retrospectively, 251 patients with cervical cancer stages IA2-IIA, who had received a radical hysterectomy, were assigned to training and validation cohorts. After the development of the scoring index using logistic coefficient analysis, the performance of the prediction model was assessed using independent validation sets. RESULTS: In the training cohort (n = 167), multivariate analysis indicated that the patient's stage, the cephalocaudal tumor diameter measured by MRI, and finding of PMI as obtained by MRI were independent predictors of PMI (P = 0.010, < 0.001, and 0.020, respectively). These predictors were internally validated by a rigorous bootstrapping method with statistical significance. The scoring index was created based on logistic coefficients, and the maximal score yielding a negative likelihood ratio less than 0.05 was selected as a cutoff. The cutoff was translated into the following criteria identifying a very low-risk group for PMI: (1) FIGO stage IA2-IB1, (2) no MRI finding suggesting PMI, and (3) cephalocaudal tumor diameter less than 1.0 cm by MRI. The negative predictive value (NPV) was 98.5% (95% confidence interval [CI]=91.7% to 100%). In the external validation cohort (n = 84), the NPV was 100% (95% CI = 90% to 100%). CONCLUSION: The current prediction model showed reliable performance for the identification of patients at low risk for PMI. It may be useful for stratification of patients and evaluation of results in future trials.
Authors: Jung-Yun Lee; Jina Youm; Jae Weon Kim; Kidong Kim; Hak Jae Kim; Jeong Yeon Cho; Min A Kim; Noh Hyun Park; Yong-Sang Song Journal: Cancer Res Treat Date: 2015-03-20 Impact factor: 4.679
Authors: Yingqiu Song; Beth Erickson; Xiaojian Chen; Guiling Li; Gang Wu; Eric Paulson; Paul Knechtges; X Allen Li Journal: Oncotarget Date: 2018-01-06