The Value Of Combining Gail Model With Ultrasonic BI-RADS Classification In Assessing Breast Cancer Risk

Objective:Breast cancer is the malignancy with highest incidence in Chinese women.Early detection and early treatment can significantly improve the total survival rate of breast cancer,but the misdiagnosis of breast cancer will also bring a heavy psychological burden to women.Therefore,there is an urgent need for getting a way to accurately assess and diagnose,especially for those who are suffering from breast diseases.Gail model and ultrasound BI-RADS classification are widely recognized and applied breast cancer assessment and diagnosis methods,but there are some limitations in both.The purpose of this study is to evaluate the diagnostic value of Gail model and ultrasound BI-RADS classification for breast cancer,and to establish a joint diagnostic model by applying binary logistic regression analysis,and finally complete the ROC curve to compare the application value of the three diagnostic methods.Methods:This study included 447 women with breast diseases who were visiting to our hospital.They were over 35 years old（Gail model can only be used to assess the risk of breast cancer in women aged 35 and over）,with clear pathological results and ultrasound diagnosis.We divided them into 155 cases of breast cancer group and 292cases of benign breast disease group according to their pathological results.As this study conducted a risk assessment for women who already have breast diseases,the Gail model was applied as a diagnostic method in this study.Firstly,the applicability of Gail model and ultrasound BI-RADS classification was evaluated respectively by calculating sensitivity,specificity,positive and negative predictive value,Yoden’s index and pathological coincidence rate.Secondly,the reasons why Gail model is not satisfied for Chinese women were discussed by analyzing the relationship between risk factors included in Gail model and breast cancer incidence,including age,and age of menarche and primipara.Finally,we combined Gail model with ultrasound BI-RADS classification to establish a prediction model by applying binary logistic regression and compared the clinical value of the three diagnostic methods through ROC curve.Results:The pathological coincidence rate,sensitivity,specificity and Yoden’s index of Gail model were 82.3%,52.3%,98.3%and 0.506 respectively.The specificity of Gail model was high and the misdiagnosis rate was low.The pathological coincidence rate,sensitivity,specificity and Yoden’s index of ultrasound BI-RADS classification were 58.4%,91.0%,41.1%and 0.321 respectively.The sensitivity was satisfied and the rate of missed diagnosis was low.Thus,the two diagnostic methods have good complementarity.In monofactor analysis,the average age of BC group was56.8±11.1 years,which was higher than that of BBD group 49.4±7.5 years（t=-8.399,P<0.001）,suggesting that age was proportional to the risk of breast cancer.However,there was no significant difference between BC group and BBD group in age of menarche and primipara（P=0.401,P=0.335）.There were 447 cases in the two groups,among which included 26 cases of family history of breast cancer,17 cases in BC group and 9 cases in BBD group.There was significant difference in family history between the two groups（P=0.002）.In group BC,18 patients had previous experience in breast biopsy,of whom 3 had experienced more than once,and 15 patients in BBD group had underwent breast biopsy before,but only one patient experienced more than once.The difference was statistically significant（P=0.031）.A joint diagnostic model was then established,P=exp(-7.912+2.756X_{Gail model}+1.176X_BI-RADS)/[1+exp(-7.912+2.756X_{Gail model}+1.176X_BI-RADS].By calculating and comparing the sensitivity and specificity of each P value,P=0.2 was finally taken as the critical point.The corresponding pathological coincidence rate was 78.5%,the sensitivity was 91.6%,the specificity was 71.9%.After plotting the ROC curve,the AUC of the joint diagnostic model,Gail model and ultrasound BI-RADS classification were 0.911（0.881-0.936）,0.825（0.787-0.859）,and 0.812（0.772-0.847）respectively.Compared with Gail model and ultrasound BI-RADS classification respectively,the combined diagnostic model has a significant difference between the two groups（P<0.001,P<0.001）.Conclusion:The accuracy of Gail model combined with ultrasound BI-RADS classification for breast cancer is superior to the two diagnostic methods alone.The joint diagnosis model has good clinical application value with its high sensitivity,specificity and discriminatory accuracy.