Skip to main navigation Skip to search Skip to main content

Response prediction after neoadjuvant chemoradiotherapy in esophageal cancer using FDG-PET and multiparametric MRI: A prospective multicenter study

  • Robin B den Boer*
  • , Stella Mook
  • , Hanneke W M van Laarhoven
  • , Francine E M Voncken
  • , Christel T Muijs
  • , Liudmila Kodach
  • , Marieke Vollebergh
  • , Jolanda M van Dieren
  • , Berthe Aleman
  • , Annemarieke Bartels-Rutten
  • , Aart Nederveen
  • , Veronique E M Mul
  • , Tom van den Ende
  • , Hildebrand Dijkstra
  • , John T M Plukker
  • , Roel J Bennink
  • , Johanna Van Sandick
  • , Gursah Kats-Ugurlu
  • , Derya Yakar
  • , Lodewijk A A Brosens
  • Astrid L H M W van Lier, Alicia S Borggreve, Ingmar L Defize, Jelle P Ruurda, Richard van Hillegersberg, Gert J Meijer*,
*Corresponding author for this work

Research output: Contribution to journalArticleAcademicpeer-review

Abstract

BACKGROUND: Accurate response assessment after neoadjuvant chemoradiotherapy (nCRT) in esophageal cancer patients could facilitate a more personalized treatment, including possible organ-preserving treatment for good responders or discontinuation of nCRT of poor responders.

PURPOSE: This multicenter study evaluated predictive value of fluorodeoxyglucose positron emission tomography with computed tomography (18FDG-PET-CT), diffusion weighted (DW) and dynamic contrast enhanced (DCE) magnetic resonance imaging (MRI) to assess response to nCRT in esophageal cancer patients.

METHODS: Between 2018 and 2022 patients scheduled to receive nCRT followed by surgery were prospectively included in 4 tertiary referral centers in The Netherlands. A 18FDG-PET-CT, DW-MRI, and DCE-MRI was performed before, during, and after nCRT. Tumor volumes were (semi-)automatically segmented and mean and max standard uptake volume (SUVmean and SUVmax) and tumor lesion glycolysis (TLG), mean and max apparent diffusion coefficient (ADC), and transfer constants Ktrans and Kep were extracted from respectively 18FDG-PET-CT, DW-MRI, and DCE-MRI scans. Absolute values and relative changes during and after nCRT were analyzed. Primary outcome was pathological complete response (pCR), defined as tumor regression grade (TRG) 1, versus non-pCR. Forward feature selection logistic regression models for pCR were developed and leave one out cross-validation (LOOCV) with bootstrapping was used to estimate performance in an 'independent' dataset.

RESULTS: 141 patients were included in the analyses of whom 33 had a pCR and 108 non-pCR. ΔTLGper, ΔSUVmax,post, and ΔADCmean,per were best univariable predictors for pCR (AUC: 0.68, 0.64, 0.62). Combining DW-MRI, DCE MRI and 18FDG-PET features resulted in a forward feature selection logistic regression model with an AUC of 0.65 (95%CI 0.48-0.79, using parameters ΔTLG,per and ΔTLG,post) after LOOCV for predicting pCR.

CONCLUSION: Despite promising results observed in previous smaller single-center studies, the predictive value of the investigated quantitative imaging parameters did not translate into clinically meaningful performance in this prospective multicenter setting.

Original languageEnglish
Article number111687
JournalRadiotherapy and oncology : journal of the European Society for Therapeutic Radiology and Oncology
Volume222
Early online date10 Jul 2026
DOIs
Publication statusPublished - Sept 2026

Fingerprint

Dive into the research topics of 'Response prediction after neoadjuvant chemoradiotherapy in esophageal cancer using FDG-PET and multiparametric MRI: A prospective multicenter study'. Together they form a unique fingerprint.

Cite this