Nomogram Predicts Overall Survival in Patients With Ovarian Cancer, Liver Metastases

A nomogram can help predict 1-, 2-, and 3-year overall survival (OS) among patients with ovarian cancer and liver metastases (OCLM), according to a study published in the World Journal of Surgical Oncology.¹

The most common site of distant metastasis from ovarian cancer is the liver.² Liver metastases occur within about 7.18% of patients with ovarian cancer, and the median OS is 11 months.³ Up to 50% of patients who died of ovarian cancer were found to have liver metastasis, meaning the actual incidence of OCLM is most likely considerably higher.⁴

Because numerous factors influence outcomes in this patient population, the researchers emphasized the difficulty of accurately predicting the survival of patients with OCLM.¹ Although prediction models are necessary, the existing ones have limitations, like inadequate sample size and the lack of external validation.

Consequently, the researchers aimed to develop and validate a more precise prognostic model to predict survival in patients with OCLM. They aimed to improve and optimize prognostic models for those with OCLM and provide clinicians with the tools to assess and manage the long-term survival of this patient population.

Researchers developed and validated a nomogram to accurately predict 1-, 2-, and 3-year overall survival (OS) in patients with ovarian cancer and liver metastases (OCLM). | Image Credit: Dr_Microbe - stock.adobe.com

To do so, the researchers extracted data from the Surveillance, Epidemiology, and End Results (SEER) database. They identified ovarian cancer cases using the International Classification of Diseases for Oncology, 3rd Edition (ICD-O-3) and site codes C56.9-Ovary.

Additionally, the researchers determined those with liver metastases as patients who answered “yes” to the “SEER Combined Mets at DX-live (2010+)” field. This determination is based on physical examination, imaging, and pathological examination, with the latter being the most critical when confirmed positive. Since there was no recording of metastases before 2010, they analyzed patients diagnosed with primary ovarian cancer between 2010 and 2021.

Consequently, the researchers identified 821 eligible patients, who they randomly divided into training (n = 574) and validation (n = 247) cohorts at a ratio of 7:3. They highlighted that there were no significant differences between the cohorts, meaning the patients were comparable.

The SEER database collected clinical patient data, including age, race, treatment, and follow-up information. The researchers assessed the clinical factors associated with OS using univariate and multivariate Cox regression analyses. Also, using the Akaike information criterion (AIC), they applied backward stepwise regression to select the optimal predictor variables.

The univariate Cox regression analysis determined that age, grade, histology, laterality, tumor size, tumor (T) stage, surgery of primary site, cytoreduction, chemotherapy, and lung, bone, and brain metastases (LBB-Met) were risk factors affecting the survival of patients with OCLM. After the backward stepwise selection based on AIC, the multivariate Cox regression analysis identified age, histology, grade, cytoreduction, chemotherapy, and LBB-Met as independent prognostic factors associated with OS.

Based on these 7 identified prognostic factors, the researchers constructed the nomogram for predicting OS in patients with OCLM. Its prediction ability was then evaluated among both cohorts using the concordance index (C-index), receiver operating characteristic (ROC) curve, calibration curve, and decision curves analysis (DCA).

In the training and validation cohorts, the C-index for the nomogram was 0.718 (95% CI, 0.691-0.745) and 0.711 (95% CI, 0.666-0.756), respectively, showing that the model was of good stability. Also, the areas under the ROC curves at 1, 2, and 3 years were 0.835, 0.748, and 0.736 in the training cohort and 0.801, 0.727, and 0.689 in the validation cohort, respectively.

Additionally, the calibration curves after 1000 bootstraps showed strong concordance between actual and predicted values in the cohorts. Lastly, they noted that DCA curves showed good positive net benefits of the nomogram at 1, 2, and 3 years. Overall, all validation results indicated that the nomogram's reliability and accuracy are satisfactory.

The researchers further evaluated the predictive ability of the nomogram by comparing it with an external model. Compared with the C-indexes among the training (0.718 [95% CI, 0.691-0.745]) and validation (0.711 [95% CI, 0.666-0.756) cohorts for the nomogram, those for the external model were 0.677 (95% CI, 0.650-0.704) and 0.643 (95% CI, 0.598-0.688) among the training and validation cohorts, respectively. Based on the time-dependent C-index curves, the nomogram was superior to the external model within 0-120 months for accuracy and stability.

Similarly, the ROC curves showed that the nomogram outperforms the external model in predicting 1, 2, and 3 years of OS. Also, the DCA results demonstrated that the nomogram’s predictive value was more significant than the external model in predicting the survival of patients with OCLM.

Lastly, the researchers calculated the net reclassification index (NRI) and the integrated discrimination improvement (IDI) to compare the models’ performances. They found that all the NRI and IDI values of the training and validation cohorts were greater than 0, further indicating the nomogram's improved predictive capacity over the external model.

The researchers concluded by acknowledging their study’s limitations, including its retrospective nature. Therefore, they cannot exclude the potential risk of selection bias. Despite their limitations, the researchers expressed confidence in their nomogram and suggested areas for further research.

“In contrast to the external model, our model shows higher stability and accuracy in predictive power and can provide a reference for clinical decision-making,” the authors concluded. “In future studies, we will further verify the clinical application value of the model.”

References

1. Xiao H, Pan N, Ruan G, Hao Q, Chen J. Development and validation of a nomogram for predicting outcomes in ovarian cancer patients with liver metastases. World J Surg Oncol. 2024;22(1):327. doi:10.1186/s12957-024-03608-x
2. Deng K, Yang C, Tan Q, et al. Sites of distant metastases and overall survival in ovarian cancer: a study of 1481 patients. Gynecol Oncol. 2018;150(3):460-465. doi:10.1016/j.ygyno.2018.06.022
3. Hou GM, Jiang C, Du JP, et al. Nomogram models for predicting risk and prognosis of newly diagnosed ovarian cancer patients with liver metastases - a large population-based real-world study. J Cancer. 2021;12(24):7255-7265. doi:10.7150/jca.64255
4. Hussain I, Xu J, Deng K, et al. The prevalence and associated factors for liver metastases, development and prognosis in newly diagnosed epithelial ovarian cancer: a large population-based study from the SEER database. J Cancer. 2020;11(16):4861-4869. doi:10.7150/jca.40590