Session: 653. Multiple Myeloma: Clinical and Epidemiological: Poster III
Hematology Disease Topics & Pathways:
Artificial intelligence (AI), Research, Clinical trials, Adult, Translational Research, Clinical Research, Patient-reported outcomes, Technology and Procedures, Study Population, Human, Imaging, Machine learning
Methods and Materials: From January 2021 to March 2024, we enrolled consecutive myeloma patients at staging into an observational prospective trial and divided them into two groups based on International Myeloma Working Group (IMWG) criteria: High-risk Smouldering Multiple Myeloma (SMM, group 1) and Multiple Myeloma (MM, group 2). All patients underwent WB-MRI using the Myeloma Response Assessment and Diagnosis System (MY-RADS). We use the term "Radiopsy" to indicate the quantification and modelling of image characteristics nearby the biopsy site to predict patient status. An experienced radiologist placed a cylindrical, 5 cc VOI nearby the biopsy site and 5 more identical VOIs on distant sites such the pelvis bone and on D11 and L5 vertebrae. The dataset was split into a training and test set with a 70:30 ratio. LASSO algorithm was used to select the most predictive features and build logistic regression models, which were then validated using the test set. Receiver operating curves (ROC) and area under the curve (AUC) were used as metrics for models’ performance assessment. Radiopsy models were tested on distant biopsy site to predict disease invasion.
Results: The study included 102 patients (46 males, mean age 63 ± 12 [SD]) with 60 diagnosed with MM and 42 with SMM. 144 quantitative features were extracted from the VOI at the biopsy site WB-MRI ADC and FF sequences for each patient. Radiopsy model showed a median AUC of 0.85 (0.79-0.95) in the training phase and a median AUC of 0.65 (0.55-0.80) in the test phase. The best predictive model had an AUC of 0.95 and 0.75 in the training and test phase, respectively. The models used to predict patient status at biopsy site were also predictive in distant VOIs.
Conclusions: Radiopsy models can distinguish between MM and SMM with good performance nearby the biopsy site. Radiopsy can be used to predict disease invasion on distant sites where biopsy is not possible or not feasable
Clinical Relevance/Application: This observational prospective trial showed that using quantitative features extracted from WB-MRI ADC and FF sequences can help distinguish between high-risk SMM and MM at staging, thus potentially enhancing diagnostic accuracy.
Disclosures: Cerchione: Beigene: Consultancy; Janssen: Consultancy; Immunogen: Consultancy; Karyopharm: Consultancy; Menarini-Stemline: Consultancy; Oncopeptides: Consultancy; Pfizer: Consultancy; Sanofi: Consultancy; Servier: Consultancy; Takeda: Consultancy; Curis: Consultancy; AMGEN: Consultancy; Abbvie, AMGEN, Astellas, Beigene, BMS, Glycomimetics, GSK, Immunogen, Janssen, Jazz, Karyopharm, Menarini - Stemline, Oncopeptides, Pfizer, Sanofi, Servier, Stemline, Takeda: Other: Advisory board; Glycomimetics: Consultancy; Astellas: Consultancy; GSK: Consultancy, Current holder of stock options in a privately-held company; Abbvie: Consultancy; Stemline: Consultancy; Skyline DX: Consultancy; Karyopharm: Consultancy; Jazz: Consultancy; GSK: Consultancy; BMS: Consultancy. Martinelli: MSD: Consultancy; ARIAD: Consultancy; Roche: Consultancy; Bristol Myers Squibb (BMS): Consultancy; Novartis: Consultancy; Pfizer: Research Funding. Normanno: MERK: Consultancy; BIOCARTIS: Consultancy; Lilly: Consultancy; ThermoFisher: Consultancy; Insight: Consultancy; AstraZeneca: Consultancy; MSD: Consultancy; ROCHE: Consultancy; QIAGEN: Consultancy; Bristol Myers Squibb (BMS): Consultancy.
See more of: Oral and Poster Abstracts