Incorporation of Tumor-Free Distance and Other Alternative Ultrasound Biomarkers into a Myometrial Invasion-Based Model Better Predicts Lymph Node Metastasis in Endometrial Cancer: Evidence and Future Prospects

Marcin Liro; Marcin Śniadecki; Ewa Wycinka; Szymon Wojtylak; Michał Brzeziński; Joanna Jastrzębska; Dariusz Wydra

doi:10.3390/diagnostics12112604

Incorporation of Tumor-Free Distance and Other Alternative Ultrasound Biomarkers into a Myometrial Invasion-Based Model Better Predicts Lymph Node Metastasis in Endometrial Cancer: Evidence and Future Prospects

Diagnostics (Basel). 2022 Oct 27;12(11):2604. doi: 10.3390/diagnostics12112604.

Authors

Marcin Liro¹, Marcin Śniadecki¹, Ewa Wycinka², Szymon Wojtylak³, Michał Brzeziński⁴, Joanna Jastrzębska¹, Dariusz Wydra¹

Affiliations

¹ Department of Gynecology and Obstetrics, Medical University of Gdańsk, 80-210 Gdańsk, Poland.
² Department of Statistics, Faculty of Management, Gdańsk University, 80-309 Gdańsk, Poland.
³ Department of Pathology, Medical University of Gdańsk, 80-210 Gdańsk, Poland.
⁴ Department of Gynecologic Oncology, PCK Marine Hospital in Gdynia, 81-519 Gdynia, Poland.

Abstract

Myometrial invasion (MI) is a parameter currently used in transvaginal ultrasound (TVS) in endometrial cancer (EC) to determine local staging; however, without molecular diagnostics, it is insufficient for the selection of high-risk cases, i.e., those with a high risk of lymph node metastases (LNM). The study’s objective was to answer the question of which TVS markers, or their combination, reflecting the molecular changes in EC, can improve the prediction of LNM. Methods: The TVS examination was performed on 116 consecutive EC patients included in this prospective study. The results from the final histopathology were a reference standard. Univariate and multivariate logistic models of analyzed TVS biomarkers (tumor [T] size, T area [AREA], T volume [SPE-VOL], MI, T-free distance to serosa [TFD], endo-myometrial irregularity, [EMIR], cervical stromal involvement, CSI) were evaluated to assess the relative accuracy of the possible LNM predictors., Spline functions were applied to avoid a potential bias in assuming linear relations between LNM and continuous predictors. Calculations were made in R using libraries splines, glmulti, and pROC. Results: LNM was found in 20 out of the 116 (17%) patients. In univariate analysis, only uMI, EMIR, uCSI and uTFD were significant predictors of LNM. The accuracy was 0.707 (AUC 0.684, 95% CI 0.568−0.801) for uMI (p < 0.01), 0.672 (AUC 0.664, 95% CI 0.547−0.781) for EMIR (p < 0.01), 0.776 (AUC 0.647, 95% CI 0.529−0.765) for uCSI (p < 0.01), and 0.638 (AUC 0.683, 95% CI 0.563−0.803) for uTFD (p < 0.05). The cut-off value for uTFD was 5.2 mm. However, AREA and VOL revealed a significant relationship by nonlinear analysis as well. Among all possible multivariate models, the one comprising interactions of splines of uTFD with uMI and splines of SPE-VOL with uCSI showed the most usefulness. Accuracy was 0.802 (AUC 0.791, 95% CI 0.673−0.91) Conclusions: A combination of uTFD for patients with uMI > 50%, and SPE-VOL for patients with uCSI, allows for the most accurate prediction of LNM in EC, rather than uMI alone.

Keywords: biomarkers; endometrial cancer; lymph nodes; metastases; model; staging; ultrasound.

Grants and funding

This research was funded by the MEDICAL UNIVERSITY OF GDANSK, grant number ST02-0074/07. The APC was funded by the Medical University of Gdansk.