Abstract 2737
Background
Well-differentiated liposarcomas (WDLPS) can be difficult to distinguish from lipomas. Currently, this distinction is made by testing for MDM2-amplification, which requires a biopsy. The aim of this study was to non-invasively predict the MDM2-amplification status using radiomics, i.e. a combination of quantitative imaging features and machine learning techniques, thereby differentiating between lipomas and WDLPS.
Methods
Patients with a lipoma or WDLPS, a known MDM2-amplification status and a pre-treatment T1-weighted MRI scan who were referred to or diagnosed at the Erasmus MC from 2009-2018 were included. When available, other sequences, e.g. T2-weighted (fat-saturated) MRI, were included in the radiomics analysis. Imaging features describing intensity, shape, orientation and texture were extracted from the tumor segmentations; age, gender, tumor depth and tumor localization were added as semantic features. Classification was performed using an ensemble of various machine learning approaches. Evaluation was performed through 100x random-split cross-validation.
Results
We included 116 patients: 58 patients with lipoma and 58 patients with WDLPS. The dataset originated from 41 MRI-scanners, resulting in large heterogeneity in imaging hardware and acquisition protocols. The best performing radiomics approach to differentiate between WDLPS and lipomas, based on T1-weighted imaging only, resulted in a mean [95% confidence interval] AUC of 0.75 [0.65-0.86], accuracy of 0.68 [0.59-0.77], sensitivity of 0.63 [0.48-0.78] and specificity of 0.73 [0.59-0.87]. A model based on the combination of imaging and semantic featured showed only a minor and non-significant improvement in performance.
Conclusions
There is a significant relationship between a radiomics signature, consisting of a combination of quantitative MRI features, and the MDM2-amplification status. Although further optimization and validation is needed for the transition to clinical practice, radiomics has appeared to be a promising, non-invasive approach for the classification of WDLPS and lipomas.
Clinical trial identification
Editorial acknowledgement
M.Vos and M.P.A. Starmans contributed equally to this study.
Legal entity responsible for the study
The authors.
Funding
Stichting Coolsingel, Netherlands Organisation for Scientific Research, Dutch Cancer Society.
Disclosure
All authors have declared no conflicts of interest.
Resources from the same session
5107 - Tolerability of PCV in Low Grade Glioma: a Real World Experience
Presenter: Razia Aslam
Session: Poster Display session 1
Resources:
Abstract
6171 - Molecular EGFR/PD-L1 profile of glioblastoma: Hints for therapeutic strategies
Presenter: Wojciech Szopa
Session: Poster Display session 1
Resources:
Abstract
4418 - Role of Intraoperative MRI (iMRI) in neurosurgery, potential imaging tool for maximizing gross total resection (GTR) rate of CNS tumours.
Presenter: Shweta Wadhwa
Session: Poster Display session 1
Resources:
Abstract
4791 - Tumor related impairments of neurocognitive functions in patients with brain metastases
Presenter: Lei Wen
Session: Poster Display session 1
Resources:
Abstract
5372 - Is radiation therapy influencing Alzheimer’s in Brain and Head and Neck Cancers?
Presenter: Anas Saad
Session: Poster Display session 1
Resources:
Abstract
2329 - Mature Treatment Results of 120 Cases of Acoustic Neuroma Treated with Cyberknife at a single institution.
Presenter: Shawn Malone
Session: Poster Display session 1
Resources:
Abstract
3207 - A phase II/III trial of high-dose, intermittent sunitinib in patients with recurrent glioblastoma: the STELLAR study
Presenter: Cyrillo Brahm
Session: Poster Display session 1
Resources:
Abstract
3589 - CNS penetration and Preliminary Efficacy of Sacutizumab Govitecan in Breast Brain Metastasis and Glioblastoma: A Surgical Study
Presenter: Andrew Brenner
Session: Poster Display session 1
Resources:
Abstract
1969 - Wedge resection with omission of lymph node dissection as an optional treatment strategy in octogenarians or older with early stage non-small cell lung cancers
Presenter: Takahiro Mimae
Session: Poster Display session 1
Resources:
Abstract
5116 - Characterization of lung tumorspheres reveals cancer stem-like cells potential targets and prognostic markers in non-small cell lung cancer
Presenter: Alejandro Herreros Pomares
Session: Poster Display session 1
Resources:
Abstract