Oops, you're using an old version of your browser so some of the features on this page may not be displaying properly.

MINIMAL Requirements: Google Chrome 24+Mozilla Firefox 20+Internet Explorer 11Opera 15–18Apple Safari 7SeaMonkey 2.15-2.23

Poster Display session

476P - How to predict microsatellite instability (MSI) status in locally advanced gastric cancer? A preoperative CT-based proposal

Date

27 Jun 2024

Session

Poster Display session

Presenters

Giulia Arrivi

Citation

Annals of Oncology (2024) 35 (suppl_1): S162-S204. 10.1016/annonc/annonc1482

Authors

G. Arrivi1, M. Polici2, E. Muttillo2, L. Di Cicco3, A. Laghi2, E. Pilozzi2, P. Mercantini2, F. Mazzuca4

Author affiliations

  • 1 Azienda Ospedaliera Sant'Andrea, Rome/IT
  • 2 Sant’ Andrea University Hospital, Rome/IT
  • 3 Sant ‘Andrea University Hospital, Rome/IT
  • 4 Azienda Ospedaliero-Universitaria Sant'Andrea - Università Sapienza di Roma, Rome/IT

Resources

Login to get immediate access to this content.

If you do not have an ESMO account, please create one for free.

Abstract 476P

Background

Microsatellite status is a prognostic and predictive biomarker in locally advanced gastric cancer (LAGC) and its testing relies on pathological exam. The aim of our study is to build a predictive model for microsatellite status, based on preoperative computed tomography (CT) scan features.

Methods

We conducted a retrospective analysis of 40 patients with LAGC, treated with surgery and adjuvant/perioperative chemotherapy at Sant’Andrea Hospital of Rome, from August 2017 to January 2024, divided into microsatellite stable (MSS) and instable (MSI) group. All patients had baseline CT scan acquired with unhenanced, late arterial, and portal venous phase. A qualitative and quantitative assessment were performed by an expert blinded radiologist. In the qualitative analysis were evaluated cTN, cEMVI status, Dmax, cancer thickness, and peritumoral fatty infiltration. In the quantitative analysis were calculated: conventional CT-HU values, and delta contrast enhancement in all CT phases. All radiological features were compared between two groups (T-test or Mann–Whitney) and their significance was tested with ROC curve analysis. All features were used to build predictive models (qualitative, quantitative, and combined model) with multivariate logistic regression.

Results

13 and 27 patients presented MSI and MSS status, respectively. In prediction of MSI status, the tumor site and cancer thickness resulted to be significant (p<0.01) with AUC=0.72 and 0.75, respectively. In the univariate analysis the same features resulted to be independently correlated with MSS/MSI status with OR ≠ 1. The combined model, including all CT scan features, achieved an AUC=0.98 (P=0.0003), with a percentage of cases correctly classified of 92.5%. The qualitative model had an AUC=0.87 (P=0.01), while the quantitative model was not significant.

Conclusions

Our non invasive radiological model predicts the MSI/MSS status with a percentage of correctly classified cases of 92.5%.It can be useful to clinicians in to classify LAGC in MSS or MSI, then to select only patients suitable for chemotherapy. In particular, the radiological model could have the most relevance when pathological tissue is lacking or in urgency for bleeding or occlusion.

Legal entity responsible for the study

Federica Mazzuca.

Funding

Has not received any funding.

Disclosure

All authors have declared no conflicts of interest.

This site uses cookies. Some of these cookies are essential, while others help us improve your experience by providing insights into how the site is being used.

For more detailed information on the cookies we use, please check our Privacy Policy.

Customise settings
  • Necessary cookies enable core functionality. The website cannot function properly without these cookies, and you can only disable them by changing your browser preferences.

  • Analytics cookies help us improve our website by collecting and reporting information on its usage.

  • We use marketing cookies to build a profile of you as a user through your actions on our site in order to better understand your interests and provide you with pertinent suggestions.