OBJECTIVE: Accurate preoperative estimates of the probability of malignancy in women with adnexal masses are essential for ensuring optimal care. This study presents a new statistical model for combining predictive information and a graphic decision support tool for calculating risk of malignancy. METHODS: The study included 153 women treated with definitive surgery for adnexal mass between 2001 and 2007 with preoperative ultrasound testing and a serum CA125. Multivariable logistic regression was used to develop a statistical model for estimating the probability of ovarian cancer as a function of age, ultrasound score, and CA125 value, with adjustments for nonlinear and interactive relationships. RESULTS: A total of 20 cases of pathologically confirmed cancer (13 invasive malignancies, and 7 tumors of low malignant potential) were identified (20/153=13%). The model obtained excellent discrimination (ROC area=0.87), explained nearly half of the observed variation in the risk of malignancy (R²=0.43), and was well calibrated across the full range of malignancy probabilities. The model equation is represented in the form of a nomogram, which can be used to calculate preoperative probability of malignancy. At a 5% risk of malignancy threshold, the model has a sensitivity of 90% and a specificity of 73%. CONCLUSIONS: Statistical models for estimating the probability of adnexal mass malignancy are substantially improved by including adjustments for non-linear relationships among key variables. A clinically relevant nomogram provides an objective tool to further aid clinicians in counseling patients and ensuring proper referral to surgical sub-specialists when indicated.
OBJECTIVE: Accurate preoperative estimates of the probability of malignancy in women with adnexal masses are essential for ensuring optimal care. This study presents a new statistical model for combining predictive information and a graphic decision support tool for calculating risk of malignancy. METHODS: The study included 153 women treated with definitive surgery for adnexal mass between 2001 and 2007 with preoperative ultrasound testing and a serum CA125. Multivariable logistic regression was used to develop a statistical model for estimating the probability of ovarian cancer as a function of age, ultrasound score, and CA125 value, with adjustments for nonlinear and interactive relationships. RESULTS: A total of 20 cases of pathologically confirmed cancer (13 invasive malignancies, and 7 tumors of low malignant potential) were identified (20/153=13%). The model obtained excellent discrimination (ROC area=0.87), explained nearly half of the observed variation in the risk of malignancy (R²=0.43), and was well calibrated across the full range of malignancy probabilities. The model equation is represented in the form of a nomogram, which can be used to calculate preoperative probability of malignancy. At a 5% risk of malignancy threshold, the model has a sensitivity of 90% and a specificity of 73%. CONCLUSIONS: Statistical models for estimating the probability of adnexal mass malignancy are substantially improved by including adjustments for non-linear relationships among key variables. A clinically relevant nomogram provides an objective tool to further aid clinicians in counseling patients and ensuring proper referral to surgical sub-specialists when indicated.
Authors: Fanny Dao; Brooke A Schlappe; Jill Tseng; Jenny Lester; Alpa M Nick; Susan K Lutgendorf; Scott McMeekin; Robert L Coleman; Kathleen N Moore; Beth Y Karlan; Anil K Sood; Douglas A Levine Journal: Gynecol Oncol Date: 2016-03-11 Impact factor: 5.482
Authors: Irene A Burger; Debra A Goldman; Hebert Alberto Vargas; Michael W Kattan; Changhon Yu; Lei Kou; Vaagn Andikyan; Dennis S Chi; Hedvig Hricak; Evis Sala Journal: Gynecol Oncol Date: 2015-06-17 Impact factor: 5.482