Clinical prediction of distant metastases in locally advanced cervical cancer

Name: Kullathorn Thephamongkhol

keyword: Cervical cancer

; Distant metastases

; Chemotherapy

; Radiotherapy

; Prediction model

Abstract: Background: Distant metastases in cervical cancer now is the leading problem and main cause of death. FIGO staging 2018 is improved but it is questionable to predict accurate distant metastases. The objective of research is firstly to build prediction model and to internally and externally validate in order to use individual prediction for big dose or more aggressive chemotherapy and secondly to quantify how many patients could be avoided aggressive chemotherapy if we accept missing one patient developing for distant metastases. Methodology: Thai cervical cancer patients from Siriraj hospital and from Ramathibodhi, Chulalongkorn and Songklanakarin are development and validation dataset respectively. Comparison of number of distant metastatic events, univariable analyses and multivariable analyses were performed to get the final model according to TRIPOD guideline recommendation. Internal validation with bootstrapping method and external validation were done to get model performance with discrimination with C statistics and calibration plot and compare with FIGO 2018 staging system. Individual prediction with beta coefficient and baseline survival was planned with weekly or monthly chemotherapy regimen to demonstrate the benefit of change of frequency of chemotherapy. Net reclassification index (NRI) was calculated to show how many patients to avoid unnecessary treatment if intention to miss 1 patient developing distant metastasis. Results: There were 397 patients in development dataset and 384 patients in validation dataset with more than 1/2 of living patients observed more than 60 months. Distant metastases were detected in 93 patients (23.4%) and 76 patients in development and validation dataset respectively. Final prediction model consists of FIGO staging, age, initial hemoglobin, histology, pelvic lymph node pathway, paraaortic lymph node level and concurrent chemotherapy frequency. New FIGO staging 2018 system with separation as stage I-II, stage III, stage III pelvic node, stage III paraaortic node, stage IV is the gold standard to be compared in terms of discrimination C statistics. C statistics of FIGO staging, our prediction model are 0.641 (95% CI 0.592-0.690), 0.690 (95%CI 0.640-0.739). After internal validation, C statistics of prediction model is 0.682. The simulation of benefit of monthly chemotherapy vs weekly chemotherapy shows more benefit in reducing distant metastases when distant metastases is very high. Calibration plot of 3 risk categories (high, intermediate, low) overlying between prediction model and Kaplan Meier method show more accuracy in all risk groups. C statistics of external validation shows 0.667 (95% CI 0.603-0.730) with no improvement from FIGO staging system ; however, for some hospital, which has the same practice pattern, we found very high C statistics as 0.751 (95% CI 0.652-0.851). Calibration plot in external validation show good overlying between prediction model and Kaplan Meier method. For NRI after internal validation, we could avoid 29 unnecessary treatments for missing 1 patient developing distant metastasis. In the same way, NRI after external validation, we could avoid 32 unnecessary treatments for missing 1 patient developing distant metastasis. Conclusion: We get moderately reliable prediction model of individual metastases prediction and good net reclassification index. This can be used for prediction for large dose chemotherapy during radiotherapy or adjuvant chemotherapy or as patient selection criteria for clinical trial of more aggressive chemotherapy in cervical cancer patients

Thammasat University. Thammasat University Library

Address: BANGKOK

Email: preserv@tu.ac.th

Name: Chamnan Tanprasertkul

Role: advisor

Name: Jayantorn Patumanond

Role: co-advisor

Name: Mantana Dhanachai

Role: co-advisor

Name: Pornpim Korpraphong

Role: co-advisor

Created: 2021

Modified: 2023-05-09

Issued: 2023-05-09

วิทยานิพนธ์/Thesis

application/pdf

DOI: 10.14457/TU.the.2021.793

eng

DegreeName: Doctor of Philosophy

Level: Doctoral Degree

Descipline: Clinical Epidemiology

Grantor: Thammasat University

©copyrights Thammasat University

ลำดับที่.	ชื่อแฟ้มข้อมูล	ขนาดแฟ้มข้อมูล	จำนวนเข้าถึง	วัน-เวลาเข้าถึงล่าสุด
1	11593kullathorn.pdf	2.22 MB

ใช้เวลา

0.041708 วินาที