Predicting cytogenetic risk in multiple myeloma using conventional whole-body MRI, spinal dynamic contrast-enhanced MRI, and spinal diffusion-weighted imaging

Abstract

Objectives: Cytogenetic abnormalities are predictors of poor prognosis in multiple myeloma (MM). This paper aims to build and validate a multiparametric conventional and functional whole-body MRI-based prediction model for cytogenetic risk classification in newly diagnosed MM.

Methods: Patients with newly diagnosed MM who underwent multiparametric conventional whole-body MRI, spinal dynamic contrast-enhanced (DCE-)MRI, spinal diffusion-weighted MRI (DWI) and had genetic analysis were retrospectively included (2011-2020/Ghent University Hospital/Belgium). Patients were stratified into standard versus intermediate/high cytogenetic risk groups. After segmentation, 303 MRI features were extracted. Univariate and model-based methods were evaluated for feature and model selection. Testing was performed using receiver operating characteristic (ROC) and precision-recall curves. Models comparing the performance for genetic risk classification of the entire MRI protocol and of all MRI sequences separately were evaluated, including all features. Four final models, including only the top three most predictive features, were evaluated.

Results: Thirty-one patients were enrolled (mean age 66 ± 7 years, 15 men, 13 intermediate-/high-risk genetics). None of the univariate models and none of the models with all features included achieved good performance. The best performing model with only the three most predictive features and including all MRI sequences reached a ROC-area-under-the-curve of 0.80 and precision-recall-area-under-the-curve of 0.79. The highest statistical performance was reached when all three MRI sequences were combined (conventional whole-body MRI + DCE-MRI + DWI). Conventional MRI always outperformed the other sequences. DCE-MRI always outperformed DWI, except for specificity.

Conclusions: A multiparametric MRI-based model has a better performance in the noninvasive prediction of high-risk cytogenetics in newly diagnosed MM than conventional MRI alone.

Critical relevance statement: An elaborate multiparametric MRI-based model performs better than conventional MRI alone for the noninvasive prediction of high-risk cytogenetics in newly diagnosed multiple myeloma; this opens opportunities to assess genetic heterogeneity thus overcoming sampling bias.

Key points: • Standard genetic techniques in multiple myeloma patients suffer from sampling bias due to tumoral heterogeneity. • Multiparametric MRI noninvasively predicts genetic risk in multiple myeloma. • Combined conventional anatomical MRI, DCE-MRI, and DWI had the highest statistical performance to predict genetic risk. • Conventional MRI alone always outperformed DCE-MRI and DWI separately to predict genetic risk. DCE-MRI alone always outperformed DWI separately, except for the parameter specificity to predict genetic risk. • This multiparametric MRI-based genetic risk prediction model opens opportunities to noninvasively assess genetic heterogeneity thereby overcoming sampling bias in predicting genetic risk in multiple myeloma.

Keywords: Diffusion magnetic resonance imaging; Genetics; Magnetic resonance imaging; Multiparametric magnetic resonance imaging; Multiple myeloma.

Fig. 1

Patient flowchart with inclusion criteria and initial retrieval, exclusion criteria and final patient selection. B12 vitamin B12, DCE dynamic contrast-enhanced, DWI diffusion-weighted imaging, EPO erythropoietin, GCSF granulocyte colony-stimulating factor, IMWG international myeloma working group, MRI magnetic resonance imaging, n number

Fig. 2

General overview of the MRI protocol and of the methods used for region-of-interest segmentation on the conventional anatomical whole-body MRI, spinal dynamic contrast-enhanced MRI, and spinal diffusion-weighted MRI sequences, for feature extraction, for feature selection, for statistical model building and for testing the models’ performances. In general, models are tested using receiver operating characteristic curve analysis including all MRI features and separate models are retested on the dataset using only the top three most predictive MRI features (in the final model with the three most prevalent features, generalizability can be reduced due to lack of external testing). AUC area-under-the-curve, b0-b1000 diffusion sensitizing gradients, DCE dynamic contrast-enhanced, DWI diffusion-weighted imaging, LASSO least absolute shrinkage and selection operator, MRI magnetic resonance imaging, NPV negative predictive value, PPV positive predictive value, PR-AUC precision-recall area-under-the-curve, ROC receiver operating characteristic, ROI region-of-interest, sens. sensitivity, spec. specificity, WB whole-body

Fig. 3

Overview of the 1.5-Tesla MRI scanning protocol using whole-body conventional anatomical MRI sequences (a–e), spinal dynamic contrast-enhanced MRI (f), and spinal diffusion-weighted imaging (g). A 77-year-old male patient with double hit high-risk IgGκ multiple myeloma with Salmon-Durie Plus and Revised International Staging System (second revision, R2-ISS) stadium II is presented, which was unresponsive to therapy and passed away 1.8 years after diagnosis. Regarding SLIM-CRAB criteria, a monoclonal bone marrow plasmacytosis of 50%, a light-chain involved/uninvolved ratio of 42, a total number of 19 focal MRI lesions > 5 mm (largest: 22 mm), a normocalcaemia, a mildly reduced renal function (glomerular filtration rate 60 mL/min/1.72 m², CKD stage G2) and a macrocytic anemia were observed. Suspected focal lesions of more than 5 mm in the 10th thoracic, 1st lumbar, and 1st sacral vertebra and the right iliac bone (white arrows) and diffuse abnormal signal intensities on all sequences are observed. A combined skeletal score of 11/13 with a combined focal and diffuse bone marrow invasion pattern can be observed. The b1000 diffusion-weighted images show severe diffusion restriction in all vertebrae and in the focal lesions of the 1st lumbar and 1st sacral vertebra (white arrows in g). The suspected focal lesion in the 10th thoracic vertebra does not show diffusion restriction (white dotted circle) or contrast enhancement, depicting its benign character due to a recent compression fracture. The spinal dynamic contrast-enhanced MRI sequence, 50 s after gadolinium contrast administration (Gadovist 7.5 mL, gadobutrol 1.0 mmol/mL, 0.1 mmol/kg, Bayer), shows intense and fast contrast uptake in the entire spine and especially in the focal lesions (white arrows in f). Cor coronal, DCE-MRI dynamic contrast-enhanced magnetic resonance imaging, DWI diffusion-weighted imaging, FS fat-saturated, Gd gadolinium, sag sagittal, STIR short tau inversion recovery, T1 T1-weighted, T2 T2-weighted

Fig. 4

Assessment of spinal dynamic contrast-enhanced MRI to obtain qualitative time-intensity curves (a), semi-quantitative (b), and quantitative (c) parametric maps and features of regions-of-interest in the spine and of reference tissues in the same patient as in Fig. 3. Cortical endplates, basivertebral veins, normal anatomical variants, and benign lesions like Schmorl’s nodules and Modic changes were avoided during segmentation. Regions-of-interest and reference tissue segmentations were matched with the anatomical sequences for optimal detailed segmentation. a Suspected focal lesions ≥ 5 mm in the 10th thoracic, 1st lumbar, and 1st sacral vertebra (arrows on the sagittal spinal dynamic contrast-enhanced MRI T1 Twist sequence, 50 s after gadolinium contrast administration) (Gadovist 7.5 mL, gadobutrol 1.0 mmol/mL, 0.1 mmol/kg, Bayer) and diffuse abnormal signal intensities can be observed in the spinal bone marrow. On the derived time-intensity curve, the thoracic and lumbar vertebral bone marrow (L3-third lumbar vertebra; T9-ninth thoracic vertebra) show active type IV curves with a steep first pass corresponding to high perfusion, high tissue vascularization, and low capillary resistance. The steep wash-in of a type IV curve and strong wash-out depict the effect of a highly vascularized region in combination with a small interstitial space. The suspected focal lesions in the 1st lumbar (FL L1) and 1st sacral (FL S1) vertebrae also show active type IV curves. The suspected focal lesion in the 10th thoracic (FL T10) vertebra shows an inactive type I curve without enhancement which is comparable to the reference paravertebral muscle vascularization, indicative of its benign character due to a recent compression fracture. Remark that the diffuse bone marrow infiltration also shows a type IV curve, indicative that active myeloma disease invades the entire spine diffusely. b Sagittal spinal positive enhancement integral parametric map generated in SyngoVia VB60 (Siemens) postprocessing software to assess the semi-quantitative features describing the time-intensity curve. Extracted features are wash-in, wash-out, arrival time, positive enhancement integral, time-to-peak, and initial area-under-the-time-intensity-curve (60 s). E.g. the positive enhancement integral is low (0.033) in the paravertebral muscles as reference tissue. The bone marrow of the ninth thoracic vertebral body and the focal lesion in the first lumbar vertebra have a positive enhancement integral of 0.244 and 0.441, respectively, which is 7–13 times higher than that of the reference muscle. c Sagittal spinal Ktrans (volume transfer constant) parametric map generated in SyngoVia VB60 (Siemens) postprocessing software to assess the quantitative features resulting from the Tofts model describing the time-concentration curve. Extracted features are Ktrans (volume transfer constant), Kep (rate constant), Ve (volume of the extracellular extravascular space), and initial area-under-the-time-concentration-curve (60 s). E.g. the Ktrans is low (0.094) in the paravertebral muscles as reference tissue. The bone marrow of the ninth thoracic vertebral body and the focal lesion in the first lumbar vertebra have a Ktrans of 1.094 and 1.494, respectively, which is 12–16 times higher than that of the reference muscle. Ao aorta, AT arrival time, A.U. arbitrary unit, DCE-MRI dynamic contrast-enhanced magnetic resonance imaging, FL focal lesion, iAUC initial area-under-the-curve, Kep rate constant, Ktrans volume transfer constant, L1/L3 first/third lumbar vertebra, PEI positive enhancement integral, s second, S1 first sacral vertebra, sag sagittal, SI signal intensity, T1 T1-weighted, T9/T10 ninth/tenth thoracic vertebra, TCC time-concentration curve, TIC time-intensity curve, TTP time-to-peak, Ve volume of the extracellular extravascular space, vs. versus

Fig. 5

Assessment of spinal diffusion-weighted imaging (a, b b1000 thoracic (a) and lumbar (b) spine images) to obtain a qualitative and (semi-)quantitative interpretation of the diffusion restriction of regions-of-interest in the spine and of reference tissues in the same patient as in Fig. 3. For the apparent diffusion coefficients and corresponding parametric maps (thoracic (c) and lumbar (d) spine), all b-values (0, 200, 400, 600, 1000) were used for analysis. Regions-of-interest and reference tissue segmentations were matched with the anatomical sequences for optimal detailed segmentation. E.g. the apparent diffusion coefficient of the ninth thoracic vertebra (diffusely invaded bone marrow), of the tenth thoracic vertebra (benign compression fracture), of the focal lesion in the first lumbar vertebra, and of the focal lesion in the first sacral vertebra (white arrows) equal 712, 1330, 801, and 658 × 10^-6 mm²/s, indicating diffusion restriction in all regions-of-interest except for the benign compression fracture in the tenth thoracic vertebra. ADC(R) apparent diffusion coefficient (ratio), bslope(R) bslope (ratio), b-value diffusion-sensitizing gradient, DWI diffusion-weighted imaging, sag sagittal, SI(R) signal intensity (ratio)

Fig. 6

Receiver operating characteristic (ROC) curves for the four final models based upon the three most frequently LASSO-selected features (= most predictive features for cytogenetic risk classification). a In the entire multiparametric MRI protocol including all sequences (conventional anatomical whole-body MRI + spinal dynamic contrast-enhanced MRI + spinal diffusion-weighted imaging). b In the conventional anatomical whole-body MRI sequence only. c In the spinal dynamic contrast-enhanced MRI sequence only. d In the spinal diffusion-weighted MRI sequence only. Overall statistical performance is expressed by the ROC-AUC (receiver operating characteristic area-under-the-curve)