Construction and validation of a nomogram model for predicting CINV in patients with gynecological malignancies
Abstract
Background
To establish a nomogram model for predicting chemotherapy-induced nausea and vomiting (CINV) in patients with gynecological malignancies based on relevant risk factors.
Methods
This retrospective study included patients with gynecological malignancies hospitalized in the oncology department of Affiliated People’s Hospital of Jiangsu University between February 2020 and October 2021. Patients were divided into a training set (between February 2020 and December 2020) and a validation set (between January 2021 and October 2021). Basic and clinical characteristics were collected and analyzed by univariate and multivariate logistic regression. A nomogram was constructed and assessed with the receiver operating characteristic curve (ROC). We have also conducted an external validation using data from 297 patients with gynecological malignancies admitted to two oncology wards at our hospital (140 patients from Ward 1 and 157 patients from Ward 2).
Results
This study comprised 148 patients in the training set and 148 in the validation set. Multivariate analysis revealed age <60 years (OR (Odds Ratio) = 4.001, 95% CI (Confidence interval) 1.349–11.872, P = 0.012), presence of motion sickness (OR = 3.841, 95% CI 1.200–12.296, P = 0.023), history of pregnancy-related vomiting (OR = 4.067, 95% CI 1.203–13.751, P = 0.024), and the use of moderate/high emetogenic chemotherapy drugs (OR = 10.299, 95% CI 2.858–37.115, P < 0.001) as independent risk factors for CINV. These factors were incorporated into a nomogram, which exhibited an area under the ROC (AUC) of 0.844, with a sensitivity of 81.4% and specificity of 80.0% at the optimal cut-off point of 159.48. The AUC for validation was 0.945, with sensitivity and specificity of 91.5% and 87.1% at the optimal cut-off point of 159.48, respectively. The external validation results showed an AUC of 0.704 (95% CI: 0.648–0.755), with a sensitivity of 93.33% and specificity of 48.15% (P = 0.001).
Conclusion
The developed nomogram, incorporating age, moderate/high emetogenic chemotherapy drugs, motion sickness, and pregnancy vomiting history, showed good discrimination for CINV.