Whole-body diffusion weighted MRI (WB-DWI) is gaining widespread recognition in initial evaluation and treatment response monitoring of bone marrow disease in multiple myeloma (MM). Apparent diffusion coefficient (ADC) has been shown to have the potential for assessing response and discriminating responder and non-responder groups in MM [1,2]. However, mono-exponential DWI model (Gaussian model), widely used in vendors’ generated ADC maps, is predisposed to fitting errors [3]. Non-Gaussian fitting algorithms might be more accurate and provide further details of microstructural changes of bone marrow [4]. This study aims to evaluate WB-DWI’s biomarkers in focal lesions (FLs) of newly diagnosed MM patients prior to (Pre) and 8 weeks (8W) following Bortezomib-based chemotherapy. Twenty-one patients (13 male, median age 52 (range 36-69)) with biopsy proven MM were enrolled prospectively. Free breathing axial diffusion weighted echo planar imaging (DWI-EPI) with spectral attenuated inversion recovery (SPAIR) plus slice selective gradient reversal (SSGR) fat suppression (TR/TE 6371/71ms, slice thickness 5mm, pixel bandwidth 3369Hz, acquisition matrix 124*72, SENSE factor 2.5, number of slices 40, b-values; 0, 100, 300 and 1000 s/mm2) was acquired from vertex to toe. Skeleton was divided to 10 anatomical locations. Images were reviewed by two radiologists in consensus and prospectively scored (0 - non-diagnostic quality images; 1=unlikely, 2=indeterminate, 3=likely and 4=highly likely disease) for the presence of focal lesions (FLs). Up to 4 largest FLs > 5mm and scored 3/4 selected for each anatomical site, to a maximum of 20 FLs per patient. A region of interest (ROI) was drawn around the focal lesion on b1000 images and then transferred to b0, 100 and 300 images (figure 1). Mean signal intensity (SI) was recorded at each b-value. Three diffusion models were used to fit the signal decay using an in-house MATLAB script: mono-exponential, stretched exponential (SE) and diffusion kurtosis models (DKI) [4, 5]. Mono-exponential diffusion coefficient (D_mono, ADC), diffusion coefficient (D_SE) and heterogeneity index (α) from SE, and diffusion coefficient (D_DKI) and diffusional kurtosis index (K) from DKI were calculated. All focal lesions were re-evaluated following treatment. Disease response was assessed by international myeloma working group (IMWG) criteria after termination of induction therapy and patients were assigned to responder and non-responder groups [6]. Difference in MRI biomarkers between Pre and 8W studies for each group was assessed by Wilcoxon test.There were 15 responders and 6 non-responders after induction chemotherapy. In total 254 focal lesions underwent quantitative biomarker analysis. The median number of lesions evaluated per patient was 14 [range 1 to 18]. The number of lesions evaluated in responders and non-responders was 186 [median 14, range1- 18] and 68 [median 12, range 7-15] respectively. There was no significant difference in per patient median lesion count between the two groups (p=0.33). D_mono increased significantly in responders (median ADC 0.80 and 1.45 x 10-3 mm2/s at Pre and 8W respectively, p=0.002) but not in non-responders (median ADC 0.61 and 0.68 x 10-3 mm2/s at baseline and early post-treatment respectively; p=0.22) (figure 2). D_SE and α both significantly increased in responders (median D_SE 0.95 and 1.49 at Pre and 8W respectively, p=0.003 and α of 0.77 and 0.81 at Pre and 8W respectively, p= 0.03) whilst no changes observed in non-responders (median D_SE 0.52 and 0.58 at Pre and 8W respectively, p=0.22 and α of 0.70 and 0.73 at Pre and 8W respectively, p= 0.31)(figure 3). There was a significant increase of D_DKI in responders (median 1.34 and 1.86 at pre and 8W respectively, p= 0.002) and a significant decrease of K (median 1.35 and 0.84 at pre and 8W respectively, p=0.02) following treatment. No significant changes of D_DKI and K were observed in non-responder group (median D_DKI and K of 1.00 and 1.94 at Pre and 1.20 and 2.04 at 8W respectively, p=0.43 and 0.56) (figure 4). In line with previous publications [1, 2] we demonstrated significant increase of Dmono in responders whilst no significant changes were observed for non-responders. Furthermore, we showed that similar pattern of temporal changes of D_SE and D_DKI occur in these groups. Observed changes of α, K in responders might be attributed to underlying accelerated destruction of the bone marrow by FLs, creating expanded spaces where behavior of water molecules move towards mono-exponential model. On the other hand, absence of any significant changes of biomarkers in non-responders could be related to a more conserved marrow structure at early phases of treatment. Longitudinal assessment of DWI related biomarkers might provide more insight to long-term changes of bone marrow following completion of treatment, remission or relapse of the disease.