Magnetic resonance imaging-based nomograms predict high-risk cytogenetic abnormalities in multiple myeloma: a two-centre study.

Objective: The study aim to use magnetic resonance imaging (MRI) radiomic features to predict high-risk cytogenetic abnormalities (HRCAs) to improve outcomes in patients with multiple myeloma (MM).

Methods: One hundred ninety-five patients with MM from two centres undergoing MRI were retrospectively recruited. Patients from Institution I (71 and 88 HRCAs and non-HRCAs, respectively) identified by fluorescence in situ hybridisation were randomly divided into training (n = 111) and validation (n = 48) cohorts. Patients from Institution II served as the external test cohort (n = 36). Radiomics or combined models based on T1WI, T2WI, and FS-T2WI images and clinical factors were constructed using logistic regression and 10-fold cross-validation in the training cohort. Nomogram performance was evaluated and compared using C-index, bootstrapping, accuracy, sensitivity, specificity, positive predictive value, negative predictive value, and Akaike information criterion. C-indexes were used to select the most efficient radiomics predictive model. Optimal model performance was tested in an external cohort.

Results: FT2+age, FT2+1+age, and FT2+2+1+age combined models were outstanding in differentiating the HRCAs of MM patients in single-, double-, and multi-sequence MRI images, respectively. The C-indexes of the training and validation cohorts corrected via the 1000 bootstrap method were 0.79 and 0.80, 0.83 and 0.84, and 0.88 and 0.84, respectively. In the external test cohort, the C-index of radiomics nomograms was 0.70, 0.76, and 0.77, respectively.

Conclusions: MRI radiomics can be used to predict HRCAs in MM patients, which will be helpful for clinical decision-making and prognosis evaluation before treatment.