Prognostic value of MRI in assessing extramural venous invasion in rectal cancer

multi-readers’ diagnostic performance

Jae Seok Bae, Se Hyung Kim, Bo Yun Hur, Won Chang, Juil Park, Hye Eun Park, Jung Ho Kim, Hyo Jin Kang, Mi Hye Yu, Joon Koo Han

Research output: Contribution to journalArticleResearchpeer-review

Abstract

Objectives: This study was conducted in order to determine the prognostic value of MRI for extramural venous invasion (EMVI) in rectal cancer compared to pathology and to assess the diagnostic performance of multireaders. Methods: We retrospectively enrolled 222 patients (M:F = 148:74; mean age ± standard deviation, 61.5 ± 12 years) with histopathologically proven rectal cancers who underwent preoperative MRI between 2007 and 2016. Among them, 74 patients had positive EMVI on pathology (pEMVI) and 148 patients had negative pEMVI. Three radiologists with 7 (reviewer 1), 3 (reviewer 2), and 1 (reviewer 3) year of experience in rectal MR imaging determined the presence of EMVI on MRI (mrEMVI) using a 5-point grading system. Using histopathologic results as the reference standard, radiologists’ performances were analyzed and compared with receiver operating characteristic (ROC) analysis. For assessment of interobserver variation, intraclass correlation coefficients (ICC) were used. Lastly, Kaplan–Meier estimation and Cox proportional hazard models were used for survival analysis. Results: The area under the ROC curve (AUC) was highest in reviewer 1 (0.829), followed by reviewer 2 (0.798) and reviewer 3 (0.658). Differences in AUCs between reviewer 1 or 2 and reviewer 3 were statistically significant (p < 0.001). ICC was substantial between reviewers 1 and 2. Overall survival (OS) was significantly different according to the positive circumferential resection margin, adjuvant treatment, and the presence of mrEMVI, but not by the presence of pEMVI. Conclusions: For experienced radiologists, the diagnostic performance of mrEMVI was good, resulting in better prediction of OS than with pEMVI, with substantial interobserver agreement. Key Points: • When read by experienced radiologists, MR can provide reliable diagnostic performance in assessing EMVI for patients with rectal cancer. • Positive mrEMVI is an adverse prognostic factor of overall survival and may influence the clinical decision-making.

Original languageEnglish
Pages (from-to)4379-4388
Number of pages10
JournalEuropean Radiology
Volume29
Issue number8
DOIs
StatePublished - 1 Aug 2019

Fingerprint

Rectal Neoplasms
Pathology
ROC Curve
Survival
Observer Variation
Survival Analysis
Proportional Hazards Models
Area Under Curve
Radiologists
Therapeutics

Keywords

  • Blood vessels
  • Chemoradiotherapy
  • Magnetic resonance imaging
  • Prognosis
  • Rectal neoplasms

Cite this

Bae, Jae Seok ; Kim, Se Hyung ; Hur, Bo Yun ; Chang, Won ; Park, Juil ; Park, Hye Eun ; Kim, Jung Ho ; Kang, Hyo Jin ; Yu, Mi Hye ; Han, Joon Koo. / Prognostic value of MRI in assessing extramural venous invasion in rectal cancer : multi-readers’ diagnostic performance. In: European Radiology. 2019 ; Vol. 29, No. 8. pp. 4379-4388.
@article{c5f789c0eab54ec4a15535449777943c,
title = "Prognostic value of MRI in assessing extramural venous invasion in rectal cancer: multi-readers’ diagnostic performance",
abstract = "Objectives: This study was conducted in order to determine the prognostic value of MRI for extramural venous invasion (EMVI) in rectal cancer compared to pathology and to assess the diagnostic performance of multireaders. Methods: We retrospectively enrolled 222 patients (M:F = 148:74; mean age ± standard deviation, 61.5 ± 12 years) with histopathologically proven rectal cancers who underwent preoperative MRI between 2007 and 2016. Among them, 74 patients had positive EMVI on pathology (pEMVI) and 148 patients had negative pEMVI. Three radiologists with 7 (reviewer 1), 3 (reviewer 2), and 1 (reviewer 3) year of experience in rectal MR imaging determined the presence of EMVI on MRI (mrEMVI) using a 5-point grading system. Using histopathologic results as the reference standard, radiologists’ performances were analyzed and compared with receiver operating characteristic (ROC) analysis. For assessment of interobserver variation, intraclass correlation coefficients (ICC) were used. Lastly, Kaplan–Meier estimation and Cox proportional hazard models were used for survival analysis. Results: The area under the ROC curve (AUC) was highest in reviewer 1 (0.829), followed by reviewer 2 (0.798) and reviewer 3 (0.658). Differences in AUCs between reviewer 1 or 2 and reviewer 3 were statistically significant (p < 0.001). ICC was substantial between reviewers 1 and 2. Overall survival (OS) was significantly different according to the positive circumferential resection margin, adjuvant treatment, and the presence of mrEMVI, but not by the presence of pEMVI. Conclusions: For experienced radiologists, the diagnostic performance of mrEMVI was good, resulting in better prediction of OS than with pEMVI, with substantial interobserver agreement. Key Points: • When read by experienced radiologists, MR can provide reliable diagnostic performance in assessing EMVI for patients with rectal cancer. • Positive mrEMVI is an adverse prognostic factor of overall survival and may influence the clinical decision-making.",
keywords = "Blood vessels, Chemoradiotherapy, Magnetic resonance imaging, Prognosis, Rectal neoplasms",
author = "Bae, {Jae Seok} and Kim, {Se Hyung} and Hur, {Bo Yun} and Won Chang and Juil Park and Park, {Hye Eun} and Kim, {Jung Ho} and Kang, {Hyo Jin} and Yu, {Mi Hye} and Han, {Joon Koo}",
year = "2019",
month = "8",
day = "1",
doi = "10.1007/s00330-018-5926-9",
language = "English",
volume = "29",
pages = "4379--4388",
journal = "European radiology",
issn = "0938-7994",
publisher = "Springer Verlag",
number = "8",

}

Prognostic value of MRI in assessing extramural venous invasion in rectal cancer : multi-readers’ diagnostic performance. / Bae, Jae Seok; Kim, Se Hyung; Hur, Bo Yun; Chang, Won; Park, Juil; Park, Hye Eun; Kim, Jung Ho; Kang, Hyo Jin; Yu, Mi Hye; Han, Joon Koo.

In: European Radiology, Vol. 29, No. 8, 01.08.2019, p. 4379-4388.

Research output: Contribution to journalArticleResearchpeer-review

TY - JOUR

T1 - Prognostic value of MRI in assessing extramural venous invasion in rectal cancer

T2 - multi-readers’ diagnostic performance

AU - Bae, Jae Seok

AU - Kim, Se Hyung

AU - Hur, Bo Yun

AU - Chang, Won

AU - Park, Juil

AU - Park, Hye Eun

AU - Kim, Jung Ho

AU - Kang, Hyo Jin

AU - Yu, Mi Hye

AU - Han, Joon Koo

PY - 2019/8/1

Y1 - 2019/8/1

N2 - Objectives: This study was conducted in order to determine the prognostic value of MRI for extramural venous invasion (EMVI) in rectal cancer compared to pathology and to assess the diagnostic performance of multireaders. Methods: We retrospectively enrolled 222 patients (M:F = 148:74; mean age ± standard deviation, 61.5 ± 12 years) with histopathologically proven rectal cancers who underwent preoperative MRI between 2007 and 2016. Among them, 74 patients had positive EMVI on pathology (pEMVI) and 148 patients had negative pEMVI. Three radiologists with 7 (reviewer 1), 3 (reviewer 2), and 1 (reviewer 3) year of experience in rectal MR imaging determined the presence of EMVI on MRI (mrEMVI) using a 5-point grading system. Using histopathologic results as the reference standard, radiologists’ performances were analyzed and compared with receiver operating characteristic (ROC) analysis. For assessment of interobserver variation, intraclass correlation coefficients (ICC) were used. Lastly, Kaplan–Meier estimation and Cox proportional hazard models were used for survival analysis. Results: The area under the ROC curve (AUC) was highest in reviewer 1 (0.829), followed by reviewer 2 (0.798) and reviewer 3 (0.658). Differences in AUCs between reviewer 1 or 2 and reviewer 3 were statistically significant (p < 0.001). ICC was substantial between reviewers 1 and 2. Overall survival (OS) was significantly different according to the positive circumferential resection margin, adjuvant treatment, and the presence of mrEMVI, but not by the presence of pEMVI. Conclusions: For experienced radiologists, the diagnostic performance of mrEMVI was good, resulting in better prediction of OS than with pEMVI, with substantial interobserver agreement. Key Points: • When read by experienced radiologists, MR can provide reliable diagnostic performance in assessing EMVI for patients with rectal cancer. • Positive mrEMVI is an adverse prognostic factor of overall survival and may influence the clinical decision-making.

AB - Objectives: This study was conducted in order to determine the prognostic value of MRI for extramural venous invasion (EMVI) in rectal cancer compared to pathology and to assess the diagnostic performance of multireaders. Methods: We retrospectively enrolled 222 patients (M:F = 148:74; mean age ± standard deviation, 61.5 ± 12 years) with histopathologically proven rectal cancers who underwent preoperative MRI between 2007 and 2016. Among them, 74 patients had positive EMVI on pathology (pEMVI) and 148 patients had negative pEMVI. Three radiologists with 7 (reviewer 1), 3 (reviewer 2), and 1 (reviewer 3) year of experience in rectal MR imaging determined the presence of EMVI on MRI (mrEMVI) using a 5-point grading system. Using histopathologic results as the reference standard, radiologists’ performances were analyzed and compared with receiver operating characteristic (ROC) analysis. For assessment of interobserver variation, intraclass correlation coefficients (ICC) were used. Lastly, Kaplan–Meier estimation and Cox proportional hazard models were used for survival analysis. Results: The area under the ROC curve (AUC) was highest in reviewer 1 (0.829), followed by reviewer 2 (0.798) and reviewer 3 (0.658). Differences in AUCs between reviewer 1 or 2 and reviewer 3 were statistically significant (p < 0.001). ICC was substantial between reviewers 1 and 2. Overall survival (OS) was significantly different according to the positive circumferential resection margin, adjuvant treatment, and the presence of mrEMVI, but not by the presence of pEMVI. Conclusions: For experienced radiologists, the diagnostic performance of mrEMVI was good, resulting in better prediction of OS than with pEMVI, with substantial interobserver agreement. Key Points: • When read by experienced radiologists, MR can provide reliable diagnostic performance in assessing EMVI for patients with rectal cancer. • Positive mrEMVI is an adverse prognostic factor of overall survival and may influence the clinical decision-making.

KW - Blood vessels

KW - Chemoradiotherapy

KW - Magnetic resonance imaging

KW - Prognosis

KW - Rectal neoplasms

UR - http://www.scopus.com/inward/record.url?scp=85059642583&partnerID=8YFLogxK

U2 - 10.1007/s00330-018-5926-9

DO - 10.1007/s00330-018-5926-9

M3 - Article

VL - 29

SP - 4379

EP - 4388

JO - European radiology

JF - European radiology

SN - 0938-7994

IS - 8

ER -