Article Text

Download PDFPDF

688 Prediction of HCC response to neoadjuvant immunotherapy using multiparametric magnetic resonance imaging: a preliminary study
  1. Octavia Bane1,
  2. Enamul Bhuiyan1,
  3. Paul Kennedy1,
  4. Muhammed Shareef1,
  5. Pauline Hamon1,
  6. Mark Buckup1,
  7. Sacha Gnjatic1,
  8. Stefanie Hectors2,
  9. Hung Kam Cheung2,
  10. Elizabeth Miller2,
  11. Maria Isabel Fiel1,
  12. Stephen Ward1,
  13. Myron Schwartz1,
  14. Thomas Marron1,
  15. Miriam Merad1 and
  16. Bachir Taouli1
  1. 1Icahn School of Medicine at Mount Sinai, New York, NY, USA
  2. 2Regeneron, Tarrytown, NY, USA


Background Despite advances in therapy, the prognosis of hepatocellular carcinoma (HCC) remains poor. The introduction of biologic drugs, including immune check-point inhibitors, has revolutionized HCC treatment. However, only a portion of patients with HCC respond to immunotherapy and predicting response is an unmet need. In this preliminary study, we assessed the value of quantitative multiparametric MRI (mpMRI) for predicting HCC response to neoadjuvant immunotherapy.

Methods In this prospective IRB-approved single-center study, we included 17 patients (M/F 14/3, mean age 65y) with resectable HCC who underwent mpMRI including T1 mapping, 3D MR elastography (MRE), diffusion-weighted imaging (DWI), and dynamic contrast-enhanced (DCE)-MRI with a hepatobiliary contrast agent (gadoxetic acid, Eovist/Primovist, Bayer), at pre-treatment and after completion of anti PD-1 immunotherapy (cemiplimab) as part of a trial. All patients underwent surgical resection. HCC lesions were identified by an experienced radiologist and regions of interest were placed to measure tumor native and post-contrast T1 measured during hepatobiliary phase (T1-HBP), tumor stiffness (TS), apparent diffusion coefficient (ADC), and perfusion parameters. The reference standard was the histopathologic percentage of necrosis. Patients with significant tumor necrosis (STN ≥50%) were considered responders. MRI parameters were compared between responders and non-responders by Mann-Whitney U test, and their diagnostic performance to predict response to treatment was assessed by ROC analysis.

Results 17 HCC lesions (6.4±4.9 cm, range 2.0-19.0 cm) were resected in 17 patients. At pre-treatment MRI, tumor native T1 and upslope from DCE-MRI were prolonged in responders compared to non-responders, and predicted response with good diagnostic performance (table 1; T1 AUC(CI)=0.82(0.58-0.99), upslope AUC(CI)=0.80(0.58-0.99). The other parameters had no value in predicting response to immunotherapy (AUC range 0.52-0.68). Tumor native T1 and T1-HBP increased with degree of tumor necrosis at resection (%).

Conclusions Our results demonstrate the potential utility of native T1 and upslope measured with DCE-MRI for predicting HCC response to neoadjuvant immunotherapy. These findings require validation in an independent study.

Acknowledgements Regeneron Pharmaceuticals, Inc. funded the study. We would like to thank our clinical coordinators, Jordan Cuevas and Jenna Korotkin for helping with patient consent and for maintaining the study database. Enamul H. Bhuiyan gratefully acknowledges Professor Steven Sourbron, The University of Sheffield for his generosity to provide updated PMI software.

Trial Registration Neoadjuvant Cemiplimab for the Treatment of Resectable NSCLC, HCC, and HNSCC identifier NCT03916627

Ethics Approval The study obtained continuing approval from the IRB at Mount Sinai through 01/19/2023, for Regeneron Pharmaceuticals, Inc. Protocol # R2810-ONC-1866/PI:

Marron/BRANY File # 19-06-061-05. All participants signed informed consent.

Abstract 688 Table 1

Comparison of mpMRI (pretreatment) parameters (tumor size, tumor native T1, tumor HBP T1, arterial plasma flow Fa; venous plasma flow Fp, total plasma flow Ft, arterial flow fraction ART, mean transit time MTT, extra cellular volume Ve, uptake fraction fi, uptake rate ki, time-to-peak TTP and upslope, apparent diffusion coefficient ADC, true diffusion coefficient D, pseudo diffusion coefficient D* and perfusion fraction PF) and their corresponding AUC for prediction of histopathologically assessed tumor response as significant tumor necrosis (STN = 50%) . Parameters with highest AUCs and significant p-values are shown in bold.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.