Differentiation of Diffuse Infiltration Pattern in Multiple Myeloma From Hyperplastic Hematopoietic Bone Marrow: Qualitative and Quantitative Analysis Using Whole‐Body MRI

• Published in: Journal of magnetic resonance imaging 2022-04, Vol.55 (4), p.1213-1225
• Main Authors: Sun, Mengtian, Cheng, Jingliang, Ren, Cuiping, Zhang, Yong, Li, Yinhua, Wang, Linlin, Liu, Yu
• Format: Article
• Language: English
• Subjects: Bone marrow; Bone Marrow - diagnostic imaging; Diffusion coefficient; Diffusion Magnetic Resonance Imaging - methods; diffusion‐weighted whole‐body imaging with background body signal suppression; Female; Field strength; Humans; hyperplastic hematopoietic bone marrow; Infiltration; Magnetic resonance imaging; Magnetic Resonance Imaging - methods; Male; modified Dixon chemical‐shift imaging; Multiple myeloma; Multiple Myeloma - diagnostic imaging; Neurotrophin 1; Neurotrophin 2; Qualitative analysis; Quantitative analysis; Retrospective Studies; Statistical analysis; Statistical tests
• ISSN: 1053-1807; 1522-2586
• DOI: 10.1002/jmri.27934

Background The visual assessment used for diffuse infiltration of multiple myeloma (MM) is inadequate. It can be difficult to differentiate MM from hyperplastic hematopoietic bone marrow (HHBM) because the MRI signal characteristics overlap. Purpose To analyze the bone marrow diffuse signal changes on whole‐body MRI caused by MM and HHBM. Study type Retrospective. Subjects Thirty Four patients with MM (21 men and 13 women), 22 patients with HHBM (9 men and 13 women), and 15 healthy controls (9 men and 6 women). Field strength/sequence A 3.0 T MRI; diffusion‐weighted whole‐body imaging with background body signal suppression (DWIBS), modified Dixon T1 fast field echo, and T2 STIR. Assessment Three radiologists analyzed the whole‐body MRI alone and in combination with apparent diffusion coefficient (ADC) and fat fraction (FF) with qualitative and quantitative analysis. Normalized T1 and T2 signal intensities (nT1 and nT2) and signal‐to‐noise ratio (SNR) were obtained. Statistical tests Kruskal–Wallis and chi‐square tests. Results The MM group had significantly higher ADC and significantly lower FF than HHBM and control groups. There was no significant difference in nT1, nT2 or SNR between MM and HHBM (P = 0.932, P = 0.097, and P = 0.110, respectively). Receiver operating characteristic (ROC) analysis using ADC and FF cut‐off values of 0.47 × 10−3 mm2/sec and 20.63%, respectively. The AUC was 0.866 for ADC and 0.886 for FF. The quantitative analysis yielded better specificity (observer 1: 81.8% vs. 27.3%; observer 2: 68.2% vs. 22.7%; and observer 3: 72.7% vs. 18.2%) and a higher diagnostic accuracy (observer 1: 82.1% vs. 51.8%; observer 2: 80.4% vs. 50.0%; observer 3: 76.8% vs. 44.6%) than the qualitative analysis. Data conclusion Whole‐body MRI combined with DWIBS and mDIXON could be used to differentiate between MM and HHBM. Combining the quantitative ADC and FF with the whole‐body MRI improved the specificity and accuracy in differentiating these conditions. Evidence Level 3 Technical Efficacy Stage 2