Evaluating the incremental value of systemic inflammatory response markers in estimating the malignant probability of ovarian masses: development of a diagnostic nomogram

Purpose: To investigate the incremental value of systemic inflammatory response markers in the differential diagnosis of ovarian masses and develop a nomogram to predict ovarian malignancy in patients with ovarian masses. Methods 1087 eligible women with ovarian masses were randomly allocated to a training cohort (N = 761) and validation cohort (N = 326). The LASSO logistic regression was implemented to select the most significant ovarian cancer-related predictors. Multivariable logistic regressions were then performed to build different prediction models. The model with best performance was presented as a nomogram and compared with reference ovarian cancer prediction tools in terms of discrimination ability. Results Seven variables including age, CA125, HE4, fibrinogen (FIB), monocyte-to-lymphocyte ratio (MLR), the presence of solid areas and strong blood flow signal were identified as significant predictors. The final model combining all seven variables yield an AUC of 0.973 in the training cohort and 0.964 in the validation cohort. The NRI and IDI revealed that the addition of FIB and MLR significantly improved the model’s discrimination ability. The nomogram based on the final model outperformed some well-known ovarian cancer prediction tools (RMI, LR2, ROMA, CPH-I and R-OPS) and exhibited an excellent performance in detecting early-stage ovarian cancer with an AUC value of 0.946 (95% CI: 0.925–0.967). Conclusion The systemic inflammatory response markers and ultrasound-based nomogram provided a convenient and accurate tool for estimating the malignant probability in women with ovarian masses. The addition of FIB and MLR significantly improved the predictive ability of the nomogram.