CC BY-NC-ND 4.0 · Laryngorhinootologie 2022; 101(S 02): S184
DOI: 10.1055/s-0042-1746471
Abstracts | DGHNOKHC
Allergology / Environmental Medicine / Immunology

PET/CT radiomics potentially improves progression-free survival (PFS) and overall survival (OS) prognostication beyond UICC TNM staging in oropharyngeal squamous cell carcinoma (OPSCC) patients

Stefan Philipp Stefan Haider
1   Yale School of Medicine, Department of Radiology and Biomedical Imaging, New Haven, CT, United States
,
Kariem Sharaf
2   Klinikum der Universität München, Klinik und Poliklinik für Hals-Nasen- Ohrenheilkunde München
,
Tal Zeevi
1   Yale School of Medicine, Department of Radiology and Biomedical Imaging, New Haven, CT, United States
,
Amit Mahajan
1   Yale School of Medicine, Department of Radiology and Biomedical Imaging, New Haven, CT, United States
,
Reza Forghani
3   McGill University Health Centre, Department of Diagnostic Radiology Montreal, Quebec, Canada
,
Benjamin L. Judson
4   Yale School of Medicine, Division of Otolaryngology, Department of Surgery New Haven, CT, United States
,
Benjamin H. Kann
5   Dana-Farber Cancer Institute, Harvard Medical School, Department of Radiation Oncology Boston, MA, United States
,
Barbara Burtness
6   Yale School of Medicine and Yale Cancer Center, Section of Medical Oncology, Department of Internal Medicine, New Haven, CT, United States
,
Christoph Reichel
2   Klinikum der Universität München, Klinik und Poliklinik für Hals-Nasen- Ohrenheilkunde München
,
Philipp Baumeister
2   Klinikum der Universität München, Klinik und Poliklinik für Hals-Nasen- Ohrenheilkunde München
,
Seyedmehdi Payabvash
1   Yale School of Medicine, Department of Radiology and Biomedical Imaging, New Haven, CT, United States
› Author Affiliations
› Further Information
Also available at
 
 

    Purpose Radiomic analysis of medical images enables automated comprehensive quantification of target lesion shape, texture, and signal intensity characteristics beyond visual assessment. We applied pre-treatment FDG-PET/non-contrast CT radiomics, UICC 8th edition TNM staging and machine learning for outcome prognostication in OPSCC.

    Methods Data from 311 OPSCC patients was retrieved from institutional archives and The Cancer Imaging Archive. Patients with known HPV-status, cM0-status at initial staging, PFS/death events or >18 months of uneventful follow-up, and treated with curative intent were included. After manual delineation of primary tumors and metastatic cervical lymph nodes, 1037 PET and 1037 CT radiomic features were extracted per lesion.

    In 33x-repeat 3-fold cross-validation, random survival forest (RSF) models for PFS and OS were trained using (1) radiomic features, (2) UICC T-, N- and overall stage concatenated with HPV-status, and (3) UICC staging combined with radiomics as input. In addition, random forest classifiers (RF) trained on radiomic features alone generated high- and low-risk groups. RSF and RF output was averaged across validation folds.

    Results In HPV+/HPV- OPSCC, RSF models for PFS yielded a mean Harrell’s C-index±SD of 0.54±0.06/0.50±0.06 (UICC), 0.62±0.05/0.55±0.07 (radiomics) and 0.62±0.05/0.56±0.07 (combined). RSF models for OS yielded 0.55±0.08/0.50±0.08 (UICC), 0.63±0.08/0.60±0.09 (radiomics) and 0.63±0.08/0.60±0.09 (combined).

    The Radiomics-based stratification of 3- to 5-year PFS and OS was significant in Kaplan-Meier analysis of HPV+ subjects, with similar trends in the smaller HPV- group.

    Conclusion PET/CT Radiomics may provide complementary value for prognostication in OPSCC.


    #

    Conflict of Interest

    The author declares that there is no conflict of interest.

    Publication History

    Article published online:
    24 May 2022

    © 2022. The Author(s). This is an open access article published by Thieme under the terms of the Creative Commons Attribution-NonDerivative-NonCommercial-License, permitting copying and reproduction so long as the original work is given appropriate credit. Contents may not be used for commercial purposes, or adapted, remixed, transformed or built upon. (https://creativecommons.org/licenses/by-nc-nd/4.0/).

    Georg Thieme Verlag
    Rüdigerstraße 14, 70469 Stuttgart, Germany