Quantitative assessment of tumor burden in multiple myeloma using MY-RADS from whole-body MRI: comparison with established prognostic biomarkers Authors: Xing Xiong, Zhi Yan, Jia Wang, Yu He, Zhenyan Zhou, Dawei Li, Yue Xiong, Changyou Zhou Published: 2026 ABSTRACT BackgroundTo explore the value of MY-RADS from whole-body MRI in assessing tumor burden in MM with reference to established prognostic biomarkers.Methods108 consecutive newly diagnosed MM patients were retrospectively enrolled. Bone marrow (BM) infiltration patterns were determined by MRI presentation. The whole-body total burden score using MY-RADS, apparent diffusion coefficient (ADC), and fat fraction (FF) of representative background BM were measured. Spearman correlation analysis was used for correlating these parameters against plasma cell infiltration ratio, and independent sample t-test was used to evaluate the differences between different International Staging System (ISS) and free light-chain ratio groups.ResultsFor diffuse, salt-and-pepper and normal patterns, patients with advanced ISS stage and high FLC-ratio had markedly lower FF and higher ADC of representative background BM. They also correlated well with plasma cell infiltration ratio. In contrast, the total burden score did not show significant difference between groups. For focal pattern, there was no significant difference of ADC and FF between groups. They didn’t correlate with plasma cell infiltration ratio as well. While, patients with advanced ISS stage and high FLC-ratio had markedly higher MY-RADS total burden score and correlated well with plasma cell infiltration ratio. For mixed pattern, ADC for L3, FF for L3 and the total burden score all showed statistical difference and correlated well with plasma cell infiltration ratio.ConclusionThe present study supports the applicability of MY-RADS as a surrogate biomarker when established prognostic biomarkers were not accessible conveniently.Supplementary InformationThe online version contains supplementary material available at https://doi.org/10.1186/s40644-026-01076-y. Introduction Multiple myeloma (MM) is a malignancy characterized by abnormal clonal plasma cell proliferation in the bone marrow (BM) [1, 2]. Tumor burden is important for treatment regimen and prognosis for MM patients [3]. It is usually assessed by the degree of plasma cell infiltration through BM biopsy and secretion of paraproteins in serum and urine known as the M-gradient at the initial diagnosis [4]. Besides, International Staging System (ISS), Durie-Salmon stage, and genetic status are also commonly used in clinic to evaluate tumor burden and risk stratification. In most cases, these tumor markers make pre-, during, and post-treatment monitoring more reliable compared with other hematologic disease [2]. Nevertheless, a minority of MM patients are non-secretory or hypo-secretory M proteins and therefore difficult to manage. [5, 6]. Besides, myeloma may escape surveillance if it expands outside the marrow cavities (extramedullary) [7, 8]. Moreover, BM biopsy is an invasive examination which is painful for patients. And the biopsy concentrates on the specific region of ilium bone which may not be representative of whole-body circumstances [9, 10]. What’s more, these clinical indicators can not reflect the anatomical heterogeneity of MM patients. Thus, seeking for a robust and noninvasive tool to assess tumor burden is important in clinical practice. Recent years, imaging is playing an important role as a surrogate to quantify the tumor burden [11]. In 1975s, Durie & Salmon criteria was initially published focusing on bone destruction for depicting tumor burden. [12] However, this criteria could only be used for indirect assessment of BM involvement based on severe bone destruction. With the advent of whole-body MRI, the excellent soft-tissue contrast allows direct visualization of BM involvement before bone destruction occurs and offers qualitative and quantitative evaluation of tumor burden of focal involvement [13]. It is generally acknowledged that patients with unequivocal focal lesions in MRI have poor outcomes [14, 15]. But the incidence of diffuse involvement is significantly higher than that of focal involvement [16]. To our knowledge, there is few objective markers in MRI examinations to determine tumor burden for diffuse involvement. In 2019, the Radiology published guidelines for acquisition, interpretation, and reporting of WB-MRI in MM, which was named Myeloma Response Assessment and Diagnosis System (MY-RADS) [17]. It recommends tumor burden assessment through total burden score, as well as apparent diffusion coefficient (ADC) and fat fraction (FF) values, regardless of focal or diffuse involvement. Since MY-RADS release, a few studies had explored its application in predicting early treatment response and prognostic role in retrospective analysis [18, 19]. However, there is few research to explore the value of MY-RADS in assessing MM tumor burden quantitatively according to different patterns. Therefore, the clinical question of this research was whether whole-body MRI parameters derived from MY-RADS (total burden score, apparent diffusion coefficient [ADC], and fat fraction [FF]) can quantitatively assess tumor burden in newly diagnose MM, using established prognostic biomarkers (ISS stage, free light-chain ratio [FLC-ratio], and bone marrow plasma cell infiltration) as reference standards. We hypothesized that these MRI parameters would correlate significantly with the prognostic biomarkers and the utility of each parameter might vary depending on the bone marrow infiltration pattern. Materials and methods Study subjects This study was approved by the Ethics Committee of the local institution and informed consent was waived. All patients with newly diagnosed MM were retrospectively included in this study from June 2019 to September 2024. Inclusion criteria: (1) Patients with MM confirmed by BM biopsy, immunofixation electrophoresis, serum protein electrophoresis and other laboratory examinations; (2) The patient underwent whole-body MRI scan and laboratory examination within one week before treatment; (3) Complete information of histological and laboratory data. Patients with a history of additional malignant tumors were excluded. Standard of reference The reference standard was the established prognostic biomarkers, including plasma cell infiltration ratio, ISS stage and FLC-ratio. ISS I and II stages were considered as the early stage whereas the ISS III stage was categorized as advanced stage [20]. FLC-ratio between 0.01 and 100 was considered as low FLC-ratio group whereas FLC-ratio > 100 or < 0.01 were divided into high FLC-ratio group [21]. The proportion of plasma cell infiltration of BM was obtained from the iliac crest [9]. The electronic medical record of each patient was reviewed for these clinical variables. MR scanning protocol All examinations were performed on a 3.0T MRI scanner (Magnetic Verio, Siemens Healthcare, Erlangen, Germany). Patients were in supine position with head first and arms placed on both sides of the body. The scanning parameters were as follows: coronal T2 TIRM sequence, repetition time (TR): 7110ms; echo time (TE): 84ms; slice thickness: 5 mm; slice spacing: 1.5 mm; FOV: 480 mm. Sagittal T1 FSE sequence, TR: 1700ms; TE: 8.6ms; slice thickness: 4 mm; slice spacing: 0.8 mm. The scan covered the skull, whole spine, pelvis and upper femur. Axial DWI sequence, TR: 4000ms; TE: 46ms; slice thickness: 5 mm; slice spacing: 1.5 mm. B values were 50s/mm2 and 800s/mm2, respectively. The scan range was the same as above. Sagittal T2 fat-suppressed (FS) sequence, TR: 3000ms; TE: 91ms; slice thickness: 4 mm; slice spacing: 0.8 mm. The sagittal Dixon parameters were as follows: dual gradient-echo 3D acquisition, TR: 3.6 ms; TE1: 1.24 ms; TE2: 2.3 ms; FOV: 448 mm; slices: 13 and water-only (W), fat-only (F) images were generated. The above scan range was T11-S1. All DWI data were transferred to Syngo MultiModality station, and the Funtool software was used to process and generate ADC images. Image analysis Two radiologists (A and B, with 5 and 7 years of experience in WB-MRI, respectively) independently analyzed all MRI images. They determined the BM infiltration pattern and calculated the MY-RADS total burden score, ADC, and FF for each patient. Both radiologists were blinded to clinical data before image review. The infiltration pattern was determined by consensus; if disagreement occurred, a third radiologist (C, with 15 years of experience) was consulted. BM infiltration patterns were classified according to the MY-RADS guidelines into the following categories [17]: (1) Normal (N): no signal abnormalities on T1, T2, or DWI; homogeneous fat signal. (2) Diffuse (D): homogeneous replacement of normal marrow signal throughout a region, appearing hypointense on T1 and hyperintense on STIR/DWI. (3) Focal (F): one or more well-defined round/oval lesions with signal characteristics different from background marrow. (4) Micronodular (also referred to as salt-and-pepper, SP): mixed punctate areas of normal and abnormal marrow, giving a mottled appearance on T1 and STIR/DWI images. This pattern corresponds to the “micronodular” pattern described in MY-RADS. (5) Focal-on-diffuse (also referred to as mixed, M): coexistence of diffuse background infiltration with superimposed focal lesions in the same patient. This pattern corresponds to the “focal-on-diffuse” pattern in MY-RADS. Patients with paramedullary or extramedullary disease were noted separately; The presence of extramedullary disease would be recorded as E+; no extramedullary disease would be recorded as E-. Following the MY‑RADS guidelines, the bone marrow was subjectively divided into seven anatomic regions: cervical spine, thoracic spine, lumbar spine, pelvis, long bones, skull, and ribs/other sites. Signal intensity changes were evaluated on diffusion‑weighted images, and focal lesion size was measured on T1 FSE and T2 TIRM sequences. According to MY-RADS, the total burden score was calculated differently by pattern: the D score was applied for N, D, and SP patterns; the F score for the F pattern; and the sum of D and F scores for the M pattern. Benign lesions and degenerative changes were not counted as focal lesions. ADC and FF measurements were performed on the representative background bone marrow of the lumbar vertebrae (L1–L5). Lumbar vertebrae were chosen because of their large size, which allows reliable ROI placement, and their frequent involvement in MM. For each vertebral body, an oval ROI (> 1 cm²) was placed on the central slice of the transverse ADC map. The same ROI was placed on sagittal water‑only Dixon images and then copied to the corresponding fat‑only Dixon images. Signal intensities from fat‑only and water‑only images were recorded, and FF= [SIF / (SIF + SIW)] ×100% (SIF, SI from the fat-only images; SIW, SI from the water-only images) (Fig. 1). ROIs were carefully positioned to avoid areas of compression fractures or focal lesions. ADC and FF values were recorded for each vertebral level from L1 to L5. The average of the two radiologists’ measurements was used for statistical analysis. Definitions of different whole-body MRI parameters derived from MY-RADS were described in Table 1. Fig. 1An example of image analysis. (A) A 62-year-old female MM patient with a diffuse MRI presentation. Whole-body MRI showed that seven anatomic regions were all infiltrated by plasma cells. The MY-RADS total burden score was ranked 7; (B, C) ROI was placed on L2 of W image manually and then copied to the corresponding location of F image. The SIF and SIW were recorded and the final mean FF was 8.2%. (D) The ROI was also set to the transverse ADC maps of L2 and it showed that the mean ADC value was 0.83 × 10− 3 mm2/s An example of image analysis. (A) A 62-year-old female MM patient with a diffuse MRI presentation. Whole-body MRI showed that seven anatomic regions were all infiltrated by plasma cells. The MY-RADS total burden score was ranked 7; (B, C) ROI was placed on L2 of W image manually and then copied to the corresponding location of F image. The SIF and SIW were recorded and the final mean FF was 8.2%. (D) The ROI was also set to the transverse ADC maps of L2 and it showed that the mean ADC value was 0.83 × 10− 3 mm2/s Table 1Definitions of different whole-body MRI parameters derived from MY-RADSParametersCalculationADCMeasured on transverse ADC maps (b = 50,800 s/mm²) using an oval ROI (> 1 cm²) placed centrally in each lumbar vertebral body (L1-L5).FFCalculated on sagittal Dixon images: FF = SIF / (SIF + SIW) × 100%, with ROI same as ADC.Total burden scoreD score1 point was assigned for each region showing diffuse diseaseF scoreThis was determined separately for each region based on both the number and the maximum size of focal lesions.Number of focal lesions: 3 points for > 10 lesions, 2 points for 2–9 lesions, 1 point for 1 lesion, 0 points for no lesion.Maximum size of the largest lesion: 3 points for > 15 mm, 2 points for 5–15 mm, 1 point for < 5 mm. Table 1Definitions of different whole-body MRI parameters derived from MY-RADSParametersCalculationADCMeasured on transverse ADC maps (b = 50,800 s/mm²) using an oval ROI (> 1 cm²) placed centrally in each lumbar vertebral body (L1-L5).FFCalculated on sagittal Dixon images: FF = SIF / (SIF + SIW) × 100%, with ROI same as ADC.Total burden scoreD score1 point was assigned for each region showing diffuse diseaseF scoreThis was determined separately for each region based on both the number and the maximum size of focal lesions.Number of focal lesions: 3 points for > 10 lesions, 2 points for 2–9 lesions, 1 point for 1 lesion, 0 points for no lesion.Maximum size of the largest lesion: 3 points for > 15 mm, 2 points for 5–15 mm, 1 point for < 5 mm. Definitions of different whole-body MRI parameters derived from MY-RADS Statistical analysis Statistical analysis was performed using SPSS (version 22.0, SPSS) and MedCalc (version 19.4.0, MedCalc Software). The normality of the data was tested using Kolmogorov-Smirnov test. Normally distributed data were expressed as mean ± standard deviation while non-normal distributed data were expressed as median (Q1, Q3). Inter-observer agreement between two radiologists of different parameters was evaluated by intraclass correlation coefficient (ICC) values. The Spearman correlation test was used to correlate ADC, FF, and total burden score against plasma cell infiltration ratio of different MM infiltration patterns, respectively. Two independent sample t-tests were performed to evaluate differences in ADC, FF, and total burden score between early/advanced ISS stages and low/high FLC-ratio groups. A P-value less than 0.05 was considered significant. Results Patient characteristics A total of 132 MM patients were admitted to hematology department at our institution. Of these, 4 patients were excluded because of additional malignant tumors. Another 20 patients were excluded because they had received regional radiation therapy. Consequently, 108 consecutive patients (62.1 ± 5.9 years, 46 female) were included, and their clinical characteristics were shown in Table 2. Table 2Patient characteristicsVariablesNumber/median (range)/mean ± SDAge (years)62.1 ± 5.9Gender (male/female)62/46Myeloma subtypes (IgG/IgA/IgD/λ-light chain/k-light chain)50/25/3/23/7ISS stage (I/II/III)18/50/40FLC-ratio (0.01–100/>100 or < 0.01)40/68Plasma cell infiltration (%)61%±24%Hemoglobin level (g/L)103.7 ± 32.9Creatinine level (µmol/L)129.2 ± 37.4Calcium level (mmol/L)2.69 (2.01, 3.10)β2 microglobulin level (mg/L)7.51 (1.88, 12.57)Albumin level (g/L)36.2 ± 8.5ISS, International Staging System; FLC-ratio, Free light-chain ratio Table 2Patient characteristicsVariablesNumber/median (range)/mean ± SDAge (years)62.1 ± 5.9Gender (male/female)62/46Myeloma subtypes (IgG/IgA/IgD/λ-light chain/k-light chain)50/25/3/23/7ISS stage (I/II/III)18/50/40FLC-ratio (0.01–100/>100 or < 0.01)40/68Plasma cell infiltration (%)61%±24%Hemoglobin level (g/L)103.7 ± 32.9Creatinine level (µmol/L)129.2 ± 37.4Calcium level (mmol/L)2.69 (2.01, 3.10)β2 microglobulin level (mg/L)7.51 (1.88, 12.57)Albumin level (g/L)36.2 ± 8.5ISS, International Staging System; FLC-ratio, Free light-chain ratio Patient characteristics ISS, International Staging System; FLC-ratio, Free light-chain ratio After evaluating by consensus, the MRI infiltration patterns of patients were classified as follows: D pattern in 33 patients, F pattern in 28 patients, SP pattern in 6 patients, M pattern in 37 patients, and N pattern in 4 patients. No patients in our cohort had paramedullary or extramedullary disease. A total of 207 sites showed diffuse disease for D pattern, 106 sites showed focal lesions for F pattern, 24 sites showed salt-and-pepper disease for SP pattern, 28 sites showed no lesion for N pattern and 189 sites showed diffuse disease, 94 sites showed focal lesions for M pattern. The ICC values of ADC, FF and MY-RADS total burden score measured by the two radiologists were 0.802 (95% CI, 0.783–0.963), 0.844 (95% CI, 0.801–0.953) and 0.853 (95% CI, 0.773–0.812), respectively, suggesting that the observed measurements of ADC and FF were highly stable. Comparison of ADC, FF and MY-RADS total burden score between early/advanced ISS stages, high/low FLC-ratio groups and correlation with plasma cell infiltration ratio for D/SP/N patterns In patients with D/SP/N patterns, the FF for L3 and ADC for L3 were significantly different between early/advanced ISS stages (p = 0.033, p<0.001) and high/low FLC-ratio groups (p = 0.007, p<0.001) (Table 3; Fig. 2). Besides, FF for L5 was also significantly different between early/advanced ISS stages (p = 0.048). Patients with advanced ISS stage and high FLC-ratio had markedly lower FF and higher ADC of representative background BM compared with early ISS stage and low FLC-ratio. FF for L3 and ADC for L3 (p<0.001, r = -0.84; p<0.001, r = 0.65) showed stronger correlations with plasma cell infiltration ratio confirmed by BM biopsy compared to other parameters (Supplement Table 1). The MY-RADS total burden score was not significantly different between groups and did not correlate with plasma cell infiltration ratio. Table 3Comparison of ADC, FF and MY-RADS total burden score between early/advanced ISS stages and high/low FLC-ratio groups for D/SP/N patternsParameterISS stage P FLC-ratio P Early(n = 23)Advanced (n = 20)Low(n = 21)High(n = 22) Total burden score 5.87 ± 0.765.58 ± 0.960.2795.58 ± 0.965.76 ± 0.700.860 FF FF-L10.22 ± 0.240.24 ± 0.130.7450.23 ± 0.160.22 ± 0.130.735FF-L20.24 ± 0.230.23 ± 0.130.8290.26 ± 0.170.22 ± 0.130.402FF-L30.30 ± 0.150.20 ± 0.110.0330.31 ± 0.150.19 ± 0.110.007FF-L40.27 ± 0.170.25 ± 0.140.7440.28 ± 0.170.24 ± 0.140.320FF-L50.31 ± 0.210.21 ± 0.130.0480.32 ± 0.170.21 ± 0.120.180FF-AV0.27 ± 0.160.23 ± 0.120.3280.28 ± 0.160.21 ± 0.120.115ADC (✕10− 6mm2/s)ADC-L1592.29 ± 97.79625.40 ± 105.370.339588.84 ± 101.85625.35 ± 99.250.285ADC-L2595.18 ± 130.81603.11 ± 96.770.832594.90 ± 137.08602.60 ± 92.210.836ADC-L3535.86 ± 101.42678.21 ± 112.28<0.001529.20 ± 101.72671.00 ± 111.12<0.001ADC-L4579.35 ± 110.55598.16 ± 100.960.571569.19 ± 107.36606.52 ± 102.690.256ADC-L5585.00 ± 121.58606.17 ± 113.560.576575.85 ± 121.77613.20 ± 112.000.319ADC-AV573.79 ± 102.49622.36 ± 100.940.132567.71 ± 104.56623.81 ± 96.890.079FF, fat fraction; ISS, International Staging System; FLC-ratio, Free light-chain ratio; AV, Average Table 3Comparison of ADC, FF and MY-RADS total burden score between early/advanced ISS stages and high/low FLC-ratio groups for D/SP/N patternsParameterISS stage P FLC-ratio P Early(n = 23)Advanced (n = 20)Low(n = 21)High(n = 22) Total burden score 5.87 ± 0.765.58 ± 0.960.2795.58 ± 0.965.76 ± 0.700.860 FF FF-L10.22 ± 0.240.24 ± 0.130.7450.23 ± 0.160.22 ± 0.130.735FF-L20.24 ± 0.230.23 ± 0.130.8290.26 ± 0.170.22 ± 0.130.402FF-L30.30 ± 0.150.20 ± 0.110.0330.31 ± 0.150.19 ± 0.110.007FF-L40.27 ± 0.170.25 ± 0.140.7440.28 ± 0.170.24 ± 0.140.320FF-L50.31 ± 0.210.21 ± 0.130.0480.32 ± 0.170.21 ± 0.120.180FF-AV0.27 ± 0.160.23 ± 0.120.3280.28 ± 0.160.21 ± 0.120.115ADC (✕10− 6mm2/s)ADC-L1592.29 ± 97.79625.40 ± 105.370.339588.84 ± 101.85625.35 ± 99.250.285ADC-L2595.18 ± 130.81603.11 ± 96.770.832594.90 ± 137.08602.60 ± 92.210.836ADC-L3535.86 ± 101.42678.21 ± 112.28<0.001529.20 ± 101.72671.00 ± 111.12<0.001ADC-L4579.35 ± 110.55598.16 ± 100.960.571569.19 ± 107.36606.52 ± 102.690.256ADC-L5585.00 ± 121.58606.17 ± 113.560.576575.85 ± 121.77613.20 ± 112.000.319ADC-AV573.79 ± 102.49622.36 ± 100.940.132567.71 ± 104.56623.81 ± 96.890.079FF, fat fraction; ISS, International Staging System; FLC-ratio, Free light-chain ratio; AV, Average Comparison of ADC, FF and MY-RADS total burden score between early/advanced ISS stages and high/low FLC-ratio groups for D/SP/N patterns FF, fat fraction; ISS, International Staging System; FLC-ratio, Free light-chain ratio; AV, Average Fig. 2Box plots of ADC for L3 (A, C) and FF for L3 (B, D) between different ISS stages and FLC-ratio groups in MM patients with D/SP/N patterns Box plots of ADC for L3 (A, C) and FF for L3 (B, D) between different ISS stages and FLC-ratio groups in MM patients with D/SP/N patterns Comparison of ADC, FF and MY-RADS total burden score between early/advanced ISS stages, high/low FLC-ratio groups and correlation with plasma cell infiltration ratio for F pattern In patients with F pattern, only MY-RADS total burden score was significantly different between early/advanced ISS stages (p = 0.006) and high/low FLC-ratio groups (p = 0.005) (Table 4; Fig. 3). It also showed significant correlation (p = 0.003, r = 0.55) with plasma cell infiltration ratio confirmed by BM biopsy (Supplement Table 2). Patients with advanced ISS stage and high FLC-ratio had markedly higher MY-RADS total burden score compared with early ISS stage and low FLC-ratio. The ADC and FF of representative background BM were not significantly different between groups and did not correlate with plasma cell infiltration ratio. Table 4Comparison of ADC, FF and MY-RADS total burden score between early/advanced ISS stages and high/low FLC-ratio groups for F patternParameterISS stage P FLC-ratio P Early(n = 16)Advanced(n = 12)Low(n = 15)High(n = 13) Total burden score 18.00 ± 6.8225.63 ± 5.920.00617.43 ± 6.0525.08 ± 6.820.005 FF FF-L10.47 ± 0.190.50 ± 0.180.6640.44 ± 0.180.53 ± 0.180.253FF-L20.51 ± 0.100.50 ± 0.110.8830.51 ± 0.100.50 ± 0.110.840FF-L30.50 ± 0.160.52 ± 0.090.6920.48 ± 0.150.54 ± 0.100.301FF-L40.54 ± 0.140.57 ± 0.110.6470.51 ± 0.120.60 ± 0.130.072FF-L50.48 ± 0.170.57 ± 0.150.1680.46 ± 0.160.58 ± 0.150.054FF-AV0.49 ± 0.140.54 ± 0.110.4230.47 ± 0.120.55 ± 0.130.115ADC (✕10− 6mm2/s)ADC-L1500.79 ± 121.16466.00 ± 181.530.572530.69 ± 139.98436.50 ± 146.890.114ADC-L2533.08 ± 93.32462.80 ± 142.380.179488.91 ± 89.23513.36 ± 149.090.646ADC-L3545.29 ± 110.50469.00 ± 127.360.123535.21 ± 126.47481.82 ± 114.590.286ADC-L4482.00 ± 104.89452.91 ± 171.570.597456.54 ± 107.85482.85 ± 160.650.628ADC-L5521.31 ± 109.73450.18 ± 168.020.194520.57 ± 122.76461.92 ± 151.870.279ADC-AV513.96 ± 79.78457.41 ± 128.760.170508.20 ± 82.68472.31 ± 124.020.381FF, fat fraction; ISS, International Staging System; FLC-ratio, Free light-chain ratio; AV, Average Table 4Comparison of ADC, FF and MY-RADS total burden score between early/advanced ISS stages and high/low FLC-ratio groups for F patternParameterISS stage P FLC-ratio P Early(n = 16)Advanced(n = 12)Low(n = 15)High(n = 13) Total burden score 18.00 ± 6.8225.63 ± 5.920.00617.43 ± 6.0525.08 ± 6.820.005 FF FF-L10.47 ± 0.190.50 ± 0.180.6640.44 ± 0.180.53 ± 0.180.253FF-L20.51 ± 0.100.50 ± 0.110.8830.51 ± 0.100.50 ± 0.110.840FF-L30.50 ± 0.160.52 ± 0.090.6920.48 ± 0.150.54 ± 0.100.301FF-L40.54 ± 0.140.57 ± 0.110.6470.51 ± 0.120.60 ± 0.130.072FF-L50.48 ± 0.170.57 ± 0.150.1680.46 ± 0.160.58 ± 0.150.054FF-AV0.49 ± 0.140.54 ± 0.110.4230.47 ± 0.120.55 ± 0.130.115ADC (✕10− 6mm2/s)ADC-L1500.79 ± 121.16466.00 ± 181.530.572530.69 ± 139.98436.50 ± 146.890.114ADC-L2533.08 ± 93.32462.80 ± 142.380.179488.91 ± 89.23513.36 ± 149.090.646ADC-L3545.29 ± 110.50469.00 ± 127.360.123535.21 ± 126.47481.82 ± 114.590.286ADC-L4482.00 ± 104.89452.91 ± 171.570.597456.54 ± 107.85482.85 ± 160.650.628ADC-L5521.31 ± 109.73450.18 ± 168.020.194520.57 ± 122.76461.92 ± 151.870.279ADC-AV513.96 ± 79.78457.41 ± 128.760.170508.20 ± 82.68472.31 ± 124.020.381FF, fat fraction; ISS, International Staging System; FLC-ratio, Free light-chain ratio; AV, Average Comparison of ADC, FF and MY-RADS total burden score between early/advanced ISS stages and high/low FLC-ratio groups for F pattern FF, fat fraction; ISS, International Staging System; FLC-ratio, Free light-chain ratio; AV, Average Fig. 3Box plots of total burden score between different ISS stages (A) and FLC-ratio groups (B) in MM patients with F pattern Box plots of total burden score between different ISS stages (A) and FLC-ratio groups (B) in MM patients with F pattern Comparison of ADC, FF and MY-RADS total burden score between early/advanced ISS stages, high/low FLC-ratio groups and correlation with plasma cell infiltration ratio for M pattern In patients with M pattern, the FF for L3, ADC for L3 and MY-RADS total burden score were all significantly different between early/advanced ISS stages (p = 0.047, p = 0.011, p = 0.021) and high/low FLC-ratio groups (p = 0.015, p = 0.004, p = 0.020) (Table 5, Supplement Fig. 1). Patients with advanced ISS stage and high FLC-ratio had markedly lower FF, higher ADC of representative background BM and higher MY-RADS total burden score compared with early ISS stage and low FLC-ratio. The FF, ADC of representative background BM and MY-RADS total burden score all showed significant correlation with plasma cell infiltration ratio confirmed by BM biopsy, with FF for L3 showing the strongest correlation compared with other parameters (Supplement Table 3). Table 5Comparison of ADC, FF and MY-RADS total burden score between early/advanced ISS stages and high/low FLC-ratio groups for M patternParameterISS stage P FLC-ratio P Early(n = 22)Advanced(n = 15)Low(n = 20)High(n = 17) Total burden score 15.05 ± 7.1320.67 ± 6.720.02114.75 ± 7.2520.35 ± 6.590.020 FF FF-L10.33 ± 0.180.28 ± 0.210.4090.35 ± 0.180.27 ± 0.200.217FF-L20.34 ± 0.190.27 ± 0.200.2620.36 ± 0.190.26 ± 0.190.124FF-L30.41 ± 0.190.27 ± 0.190.0470.42 ± 0.180.27 ± 0.180.015FF-L40.39 ± 0.190.34 ± 0.250.4470.42 ± 0.170.31 ± 0.240.132FF-L50.41 ± 0.190.33 ± 0.230.2210.43 ± 0.190.32 ± 0.220.120FF-AV0.38 ± 0.180.30 ± 0.210.2120.40 ± 0.180.29 ± 0.200.083ADC (✕10− 6mm2/s)ADC-L1615.23 ± 157.19645.13 ± 191.310.606614.20 ± 164.13642.82 ± 180.260.616ADC-L2603.27 ± 170.03645.73 ± 173.740.465600.60 ± 178.52643.88 ± 162.610.449ADC-L3548.18 ± 112.96662.07 ± 138.230.011536.75 ± 109.65662.13 ± 131.450.004ADC-L4592.41 ± 200.97650.13 ± 220.550.415557.55 ± 129.53684.35 ± 261.290.064ADC-L5596.50 ± 187.86653.53 ± 224.660.408591.10 ± 195.15653.18 ± 211.790.360ADC-AV591.12 ± 144.54650.90 ± 164.110.250580.04 ± 142.51656.90 ± 159.590.131FF, fat fraction; ISS, International Staging System; FLC-ratio, Free light-chain ratio; AV, Average Table 5Comparison of ADC, FF and MY-RADS total burden score between early/advanced ISS stages and high/low FLC-ratio groups for M patternParameterISS stage P FLC-ratio P Early(n = 22)Advanced(n = 15)Low(n = 20)High(n = 17) Total burden score 15.05 ± 7.1320.67 ± 6.720.02114.75 ± 7.2520.35 ± 6.590.020 FF FF-L10.33 ± 0.180.28 ± 0.210.4090.35 ± 0.180.27 ± 0.200.217FF-L20.34 ± 0.190.27 ± 0.200.2620.36 ± 0.190.26 ± 0.190.124FF-L30.41 ± 0.190.27 ± 0.190.0470.42 ± 0.180.27 ± 0.180.015FF-L40.39 ± 0.190.34 ± 0.250.4470.42 ± 0.170.31 ± 0.240.132FF-L50.41 ± 0.190.33 ± 0.230.2210.43 ± 0.190.32 ± 0.220.120FF-AV0.38 ± 0.180.30 ± 0.210.2120.40 ± 0.180.29 ± 0.200.083ADC (✕10− 6mm2/s)ADC-L1615.23 ± 157.19645.13 ± 191.310.606614.20 ± 164.13642.82 ± 180.260.616ADC-L2603.27 ± 170.03645.73 ± 173.740.465600.60 ± 178.52643.88 ± 162.610.449ADC-L3548.18 ± 112.96662.07 ± 138.230.011536.75 ± 109.65662.13 ± 131.450.004ADC-L4592.41 ± 200.97650.13 ± 220.550.415557.55 ± 129.53684.35 ± 261.290.064ADC-L5596.50 ± 187.86653.53 ± 224.660.408591.10 ± 195.15653.18 ± 211.790.360ADC-AV591.12 ± 144.54650.90 ± 164.110.250580.04 ± 142.51656.90 ± 159.590.131FF, fat fraction; ISS, International Staging System; FLC-ratio, Free light-chain ratio; AV, Average Comparison of ADC, FF and MY-RADS total burden score between early/advanced ISS stages and high/low FLC-ratio groups for M pattern FF, fat fraction; ISS, International Staging System; FLC-ratio, Free light-chain ratio; AV, Average Discussion In this study, we examined the role of WB-MRI parameters derived from MY-RADS in evaluating tumor burden for newly diagnosed MM patients, analyzing the correlation between ADC, FF, total burden score, and established clinical prognostic biomarkers across different BM infiltration patterns. Our results demonstrate that the optimal imaging parameter for assessing tumor burden depends on the underlying infiltration pattern: quantitative ADC and FF values of representative background BM (particularly at L3) are informative for D, SP, and N patterns, whereas the MY-RADS total burden score provides more relevant information for F pattern. For M pattern, both types of parameters contribute complementary information. These findings support our hypothesis that the utility of MY-RADS parameters varies by infiltration pattern and highlight the importance of pattern-specific interpretation of WB-MRI data in MM. The ISS stage is a simple risk stratification method based on the serum β2-microglobulin and albumin. These biomarkers suggest a high proliferation rate of tumors and have been recognized as prognostic factors in MM [20]. According to survival duration, ISS I and II stage were identified as early stage and ISS III stage was grouped into advanced stage. However, the β2-microglobulin level is strongly dependent on renal function, so it is not a specific indicator of tumor burden in MM [22]. Since different serum paraproteins of the myeloma subtypes (e.g., IgA, IgG) have different thresholds, we only considered the serum FLC-ratio into consideration which is also an important indicator of myeloma activity [3]. In 2014, the International Myeloma Working Group included the serum FLC-ratio in diagnosing symptomatic MM, and defined > 100 or < 0.01 as biomarkers for ultra high-risk smoldering MM patients [21]. Therefore, we used threshold value of 100 and 0.01 to divide patients into low and high FLC-ratio groups. FF quantification is a useful MRI technique for detecting myeloma infiltration. It is increasing with age and could reach up to 70% around the age of 60 due to gradual replacement of hematopoietic marrow by adipose cells [23]. When myelomatous infiltration occurs, marrow fat content can reduce sharply [24]. Therefore, FF quantification may potentially be used as a noninvasive tool for assessing tumor burden in MM patients. In our study, FF for L3 of advanced ISS stage and high FLC-ratio group was lower than that of early ISS stage and low FLC-ratio group for D/SP/N/M patterns and correlated well with plasma cell infiltration ratio. It could be properly explained that advanced ISS stage and high FLC-ratio group mean more myelomatous infiltration replaced fat cell, so FF was lower than that of early ISS stage and low FLC-ratio group and negatively correlated with plasma cell infiltration ratio. ADC is another promising MRI-based marker in patients with monoclonal plasma cell disease [25]. The infiltrated BM shows as increased SI on b-value images and high ADC values compared to the normal background BM due to the high cellularity, lack of fat component and destruction of trabecular bone [26]. Sun et al. conducted a study to assess the tumor burden in MM patients using whole-body based ADC and FF to identify the correlation between these parameters and prognostic biomarkers. They found that ADC and FF values significantly differed between different DS, ISS, and R-ISS stages and changed as the stage progression in the normal/diffuse infiltration pattern, but not in the focal pattern [27]. However, the selection of their representative region is time-consuming and not reproducible. In our study, we have chosen lumbar vertebral as the representative region due to the large size of those vertebrae with maximized reliable measurement, typical sites of BM infiltration and minimally affected by the intrauterine device [28]. Among the lumbar vertebrae, L3 showed the most consistent statistical significance across patterns, which may be attributed to its distinct biomechanical and physiological properties [29]. Compared with adjacent vertebrae, L3 experiences greater axial mechanical stress and local ischemia, resulting in a higher proportion of adipogenesis and lower hematopoietic marrow content under physiological conditions. Consequently, when plasma cells infiltrate the marrow, the relative change in signal intensity at L3 may be more pronounced and more readily detectable than at other lumbar levels. Importantly, however, we do not claim that L3 is the sole representative vertebra; rather, our findings suggest that L3 may be a particularly sensitive site for detecting diffuse marrow infiltration, and future studies using whole-spine volumetric segmentation may provide more comprehensive assessment. Besides, we concentrated on the MY-RADS total burden score which is a novel imaging parameter proposed by radiology consensus. It is comprehensive in considering both focal and diffuse disease, whole-body anatomic sites, as well as both the number and size of focal lesions. As we can see, it was important for F pattern in indicating the tumor burden. The neoplastic plasma cells tend to form different clusters, which could be small or large for F pattern [30]. The heterogeneity of clusters may be the reason for the L1-L5 not representative to reflect the whole-body circumstances. Earlier studies had also proposed some scoring methods for BM involvement in MM [31–33. However, these methods have been limited by focusing on only focal lesions or lesions in certain regions. This study has several limitations. First, the FF measurement using the Dixon sequence in our study was based on a dual-echo method to separate water and fat signal intensities with symmetrical sampling, which may be ambiguous (water–fat swaps) due to B0 inhomogeneity. Future studies using multi-echo Dixon techniques with B0 correction may provide more robust FF quantification. Second, the representative background BM was defined as L1–L5. Since skeletal involvement in MM is highly heterogeneous, the use of specific vertebral bodies might not be the most appropriate approach for assessing whole-body circumstances. In the future, mature AI-based segmentation techniques may be employed to provide automated whole-spine volumetric analysis, thereby overcoming this limitation. Third, due to the low incidence, there were no cases of extramedullary disease in our study. We should acknowledge that this as an area for future investigation, as extramedullary disease represents an important manifestation of MM that warrants dedicated study. Fourth, although L3 showed the most consistent statistical significance, the lack of significant findings at other vertebral levels should be interpreted with caution, and the total burden score remains a valuable complementary parameter, particularly for F and M patterns. Fifth, given the complexity of infiltration patterns and the diversity of parameters, we did not construct predictive models integrating the metrics evaluated herein; this study should therefore be considered preliminary, and further research incorporating multivariable models is warranted. Finally, we correlated tumor burden with clinically established prognostic factors, which represent indirect endpoints related to survival; future studies with longitudinal follow-up and direct survival analysis are needed to validate the prognostic value of these imaging parameters. Conclusions An integrated approach that combines pattern-specific assessment using both the total burden score and regional ADC/FF measurements may serve as supplemental imaging biomarkers of tumor burden in MM. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 1 Supplementary Material 2 Supplementary Material 2 Supplementary Material 3 Supplementary Material 3