Использование WB-LDCT для проведения различия между доброкачественными и злокачественными повреждениями в аппендикулярном скелете пациентов с заболеваниями плазменных клеток.

1. Introduction

Plasma cell diseases are malignant hematological disorders characterized by abnormal proliferation of monoclonal plasma cells, which are divided into 3 stages: monoclonal gammopathy of undetermined significance (MGUS), smoldering multiple myeloma (SMM), and multiple myeloma (MM). [ 1 – 4 ] Whole-body low-dose computed tomography (WB-LDCT) is a reasonable and cost-effective initial imaging approach. Whole-body magnetic resonance imaging (WB-MRI) is the most sensitive technique for detecting bone involvement and assessing painful complications. Positron emission tomography (PET)/computed tomography (CT) is the best tool for evaluating treatment response. [ 4 ] WB-LDCT represents the primary screening method for plasma cell diseases as it enables the detection of bone damage as well as soft tissue tumors (e.g., medullary lesions of the appendicular skeleton [AS; humerus/femur] and the adjacent bone), with plasma cell tumors (PCTs) lying adjacent to the bone being easily detectable due to bone destruction. [ 5 ] However, distinguishing between malignant PCTs and benign red bone marrow (RBM) within the medullary cavity can be challenging. [ 5 – 7 ] Several studies to date, including those conducted by Koutoulidis et al, have reported that medullary lesions in the AS of plasma cell disease patients indicate increased tumor burden and, consequently, decreased 5-year progression-free survival and overall survival. [ 8 – 14 ] Furthermore, medullary lesions in the AS also promote the progression of MM in patients with MGUS and SMM. However, these studies classified medullary lesions as a whole and did not distinguish between benign and malignant tumors. [ 8 – 14 ]

Pathological examination of lesions within the medullary cavities of the AS can be challenging because of the difficulties associated with obtaining samples from the cortical bones in this region, which are not easily accessible without causing significant trauma. Therefore, WB-MRI remains the primary diagnostic tool despite the associated disadvantages of increased cost, the need for advanced equipment and technical expertise, low detection rates, prolonged examination time (45 minutes to 1 hour), and inapplicability to patients with contraindications. [ 2 , 3 , 15 , 16 ] Thus, there is an imperative clinical need for the application of WB-LDCT to differentiate between benign and malignant lesions, which can facilitate the imaging diagnosis of MGUS/SMM/MM by using the SLiM (Sixty, light chain, and MRI) standards (M), particularly in patients without bone destruction and with a lower tumor burden. [ 1 , 2 ]

Accordingly, the current study aimed to examine the efficacy of WB-LDCT and various clinical parameters evaluated using receiver operating characteristic (ROC) curves to differentiate between benign and malignant lesions within the AS medullary cavities.

2. Materials and methods

2.1. Patient data

Informed consent was obtained from all patients. This study was approved by the hospital’s institutional review board (IIT2020023-EC-1). All relevant data are within the paper and its Supporting Information files. The clinical and radiological data of all patients with plasma cell disease admitted to our hospital between December 2020 and February 2023 were retrospectively analyzed. The study inclusion criteria were as follows: patients diagnosed with plasma cell diseases for the first time according to guidelines, [ 1 ] with myeloma-related osteolyses appearing as focal destructive lesions of the cancellous (trabecular) bone with a diameter of 5 mm or more, without a sclerotic border (unless previously treated) [ 1 ] ; patients who underwent WB-LDCT and WB-MRI examinations before treatment and at an interval ≤7 days; and those exhibiting soft tissue masses with a CT value ≥−30 hounsfield unit (HU) in the AS marrow cavities (humerus and/or femur) on WB-LDCT examination. The exclusion criteria were as follows: patients exhibiting involuntary movements, foreign bodies, or metal implants causing significant image artifacts; those with lesions exhibiting calcification; and/or those with incomplete patient data or a lack of follow-up WB-MRI images.

The patients’ clinical data, including name, sex, age, white blood cell count, hemoglobin level, platelet count, uric acid, urea, creatinine, blood calcium, lactate dehydrogenase (LDH), albumin, blood β2-microglobulin (B2M), blood M protein, urine M protein, bone marrow plasma-cell percentage (BMPC%; diagnosed by pathology), quantitative kappa (κ) light chain, quantitative lambda (λ) light chain, and free light chain (κ/λ) ratio, were recorded.

The radiological data collected were as follows: WB-LDCT included lesion morphology, location, extent of humeral/femoral involvement, area, CT value, and boundaries; and WB-MRI included diffusion-weighted imaging (DWI) signal intensity of lesions, apparent diffusion coefficient (ADC) value, T1-weighted imaging (T1WI) in-phase (IP) signal, and T1WI out-of-phase (OP) signal at initial diagnosis and after 4 treatment cycles.

2.2. Examination methods

WB-LDCT examination was performed using a uCT 960 + 320-row CT scanner (United Imaging, China). The patients were placed in a supine position with their hands secured above their heads. The scanning range extended from the elbows to the knees, and the scan parameters used were as follows: tube voltage, 120 kV; tube current, 70 mA; collimation, 0.625; matrix size, 512 × 512; iterative reconstruction (iDose4) level 2; initial slice thickness, 5 mm; subsequent slice thickness, 1 mm; and interslice spacing, 0.5 mm. Coronal multiplanar reconstruction (MPR) of the entire spine and long-axis coronal MPR of the bilateral humeri and femurs were performed, with both sets of images rendered in bone and soft tissue windows.

WB-MRI was performed using a Skyra 3.0T superconducting MRI system (Siemens Healthcare, Germany). The patient’s hands were secured at their sides with support. The scanning range extended from the top of the skull to the knees, and a combination of head and neck coils and body coils (n = 3) was used to divide the entire body into 5 segments for imaging using stitching techniques. The scanning sequences were as follows: axial DWI and ADC sequences ( b value: 50/800, repetition time [TR]: 7150 ms, echo time [TE]: 63 ms, and slice thickness: 5 mm); PET-like sequences generated by simultaneous initiation of maximum intensity projection and grayscale inversion; axial T2-weighted sequences (TR: 6940 ms, TE: 78 ms, and slice thickness: 5 mm); axial T1-Dixon sequences (TR: 3.9 ms, TE: 1.2 ms, and slice thickness: 5 mm); sagittal T1-weighted sequences (TR: 600 ms, TE: 8.5 ms, and slice thickness: 5 mm); and sagittal T2-Dixon sequences (TR: 2500 ms, TE: 86 ms, and slice thickness: 5 mm).

2.3. Image processing and measurement methods

One radiologist (LY) reviewed the MPR images of the bilateral humeri and femurs in the soft tissue window of all included patients, identified, and numbered each patient’s lesions with diameters ≥10 mm. Each patient randomly selected 1 lesion according to the lesion number as the target lesion for inclusion, using the random number table method.

WB-LDCT: For all target lesions included, regions of interest (ROIs) covering at least 80% of the lesion area were manually outlined using a polygonal tool along the edge of the lesion by ZKM, carefully avoiding the cortical bone, calcified plaques, and surrounding yellow bone marrow. Each lesion was assigned a unique identifier, and its morphology (i.e., nodular: long axis < twice the short axis; cast type: long axis > twice the short axis), location (i.e., humerus or femur), extent of humeral/femoral involvement (i.e., within or beyond the proximal half of the humerus/femur), boundaries (i.e., well-defined or poorly defined), and area and CT values (i.e., the ROI of each lesion was measured using the workstation polygonal tool) were recorded (see Fig. 1 ).

Example diagrams of medullary lesion types and measurement methods in the left femur. Normal bone marrow, CT value −56 HU (A); abnormal bone marrow (nodular), CT value 66 HU (B); abnormal bone marrow (nodular), CT value 10 HU (C); abnormal bone marrow (cast type), CT value 92 HU (D); abnormal bone marrow (cast type), CT value 17 HU (E). CT = computed tomography, HU = hounsfield unit.

WB-MRI (at initial diagnosis and after 4 treatment cycles): The ROIs of the included target lesions were outlined using a polygonal tool along the edges of each lesion by ZKM such that they covered at least 80% of the lesion area. The DWI signal intensity, ADC value, T1WI-IP signal ( S IP ), and T1WI-OP signal ( S OP ) of each lesion were measured and recorded to calculate the signal suppression ratio (SR) using the following formula:

The DWI signal intensity observed upon comparing the lesions and adjacent muscles was categorized as higher, constant, or lower, whereas the ADC value, S IP , and S OP of the corresponding sequences were measured by outlining the ROIs in the lesions using the workstation polygonal tool (see Fig. 2 ).

Sequences of WB-MRI. T1WI-IP sequences (A), T1WI-OP sequences (B), DWI sequences (C), and ADC sequences (D). ADC = apparent diffusion coefficient, DWI = diffusion-weighted imaging, IP = in-phase, OP = out-of-phase, T1WI = T1-weighted imaging, WB-MRI = whole-body magnetic resonance imaging.

2.4. Diagnostic criteria for PCT lesions

PCT lesions were identified by an associate chief physician (YL) using the following diagnostic criteria: high signal intensity on DWI sequences; S SR ≤20%; ADC value ranging from 0.5 to 1.2 × 10 3 mm 2 /s; and follow-up WB-MRI after 4 treatment cycles showing an ADC value ≥1.2 × 10 3 mm 2 /s or signal changes in the lesion conforming to those of PCTs at other locations [ 4 , 15 ] (see Figs. 3 – 6 ).

Red bone marrow (nodular). WB-LDCT showed a lesion in the right femur (A) (arrows). T1WI-IP sequences (B), S IP  = 290; T1WI-OP sequences (C), S OP  = 3; S SR  = (290 − 3)/290 = 99%. High signal intensity on DWI sequences (D). ADC value 455 × 10 3 mm 2 /s (E). Mild concentration on PET-like sequences (F). ADC = apparent diffusion coefficient, DWI = diffusion-weighted imaging, IP = in-phase, OP = out-of-phase, PET = positron emission tomography, SR = suppression ratio, T1WI = T1-weighted imaging, WB-LDCT = whole-body low-dose computed tomography.

Plasma cell tumor (cast type). WB-LDCT showed a lesion in the right femur (A) (transverse line). T1WI-IP sequences (B), S IP  = 210; T1WI-OP sequences (C), S OP  = 194; S SR  = (210 − 194)/210 = 7.6%. High signal intensity on DWI sequences (D). ADC value 973 × 10 3 mm 2 /s (E). Apparent concentration on PET-like sequences (F). ADC = apparent diffusion coefficient, DWI = diffusion-weighted imaging, IP = in-phase, OP = out-of-phase, PET = positron emission tomography, SR = suppression ratio, T1WI = T1-weighted imaging, WB-LDCT = whole-body low-dose computed tomography.

Plasma cell tumor (nodular). WB-LDCT showed a lesion in the right femur (A) (arrows). T1WI-IP sequences (B), S IP  = 250; T1WI-OP sequences (C), S OP  = 223; S SR  = (250 − 223)/250 = 10.8%. High signal intensity on DWI sequences (D). ADC value 786 × 10 3 mm 2 /s (E). Apparent concentration on PET-like sequences (F). ADC = apparent diffusion coefficient, DWI = diffusion-weighted imaging, IP = in-phase, OP = out-of-phase, PET = positron emission tomography, SR = suppression ratio, T1WI = T1-weighted imaging, WB-LDCT = whole-body low-dose computed tomography.

Red bone marrow (cast type). WB-LDCT showed a lesion in the right femur (A) (transverse line). T1WI-IP sequences (B), S IP  = 235; T1WI-OP sequences (C), S OP  = 27; S SR  = (235 − 27)/235 = 88.5%. High signal intensity on DWI sequences (D). ADC value 405 × 10 3 mm 2 /s (E). Mild concentration on PET-like sequences (F). ADC = apparent diffusion coefficient, DWI = diffusion-weighted imaging, IP = in-phase, OP = out-of-phase, PET = positron emission tomography, SR = suppression ratio, T1WI = T1-weighted imaging, WB-LDCT = whole-body low-dose computed tomography.

2.5. Statistical analysis

All statistical analyses were performed using SPSS 22.0 (International Business Machines Corporation). Normally distributed measurement data are expressed as mean ± standard deviation ( x̄ ± s ), and intergroup comparisons were carried out using independent-samples t tests. Variables that did not exhibit a normal distribution were reported as median and first and third quartiles ( m [ q 1 , q 3 ]), and intergroup comparisons were performed using the Mann–Whitney U test. Enumeration data were reported as n (%), and intergroup comparisons were performed using the chi-square test. The diagnostic value of each parameter was evaluated using ROC curves, with an area under the curve (AUC) exceeding 0.9 indicating excellent diagnostic efficacy. The level of statistical significance was set at P  < .05.

3. Results

3.1. Patient demographics

The study sample included 234 patients (males: 128; females: 106) with a mean age of 59.5 ± 8.4 (range: 41–78) years. The proportions of MGUS, SMM, and MM were 14/234, 35/234, and 185/234, respectively (Table 1 ).

Table 1 Patient demographics. Patient (n = 234) Age (yr) 59.5 ± 8.4 Gender (%)  Male 128 (54.7)  Female 106 (45.3) White blood cell (10 9 /L) 5.8 ± 4.4 Platelet (10 9 /L) 169.7 ± 64.1 Hemoglobin (g/L) 95.6 ± 21.2 Uric acid (μmol/L) 476.3 ± 165.3 Urea (mmol/mL) 8.0 ± 5.1 Creatinine (μmol/L) 135.7 ± 130.3 Blood calcium (mmol/L) 2.3 ± 0.2 LDH (U/L) 182.2 ± 72.9 Albumin (g/L) 35.1 ± 8.5 Blood β2-microglobulin (mg/L) 8.2 ± 6.2 Urine M protein (mg/24 h) 2.9 ± 3.6 BMPC (%) 56.2 ± 8.0 Blood M protein (g/L) 34.8 ± 32.3 Free light chain ratio (κ/λ) 1.67 (0.13, 23.2) κ (mg/L) 550.0 (233.5, 3660.0) λ (mg/L) 501.0 (150.0, 2220.0) BMPC = bone marrow plasma-cell percentage, LDH = lactate dehydrogenase.

Patient demographics.

BMPC = bone marrow plasma-cell percentage, LDH = lactate dehydrogenase.

3.2. Comparison of clinical data

After completion of image processing and measurement, a total of 234 lesions (1 lesion per patient) were included in the study, and their corresponding clinical parameters were matched. The lesions were categorized into PCT (n = 154) and RBM (n = 80) groups according to the PCT diagnostic criteria.

The PCT group exhibited significantly lower platelet counts (149 × 10 9 /L vs 192 × 10 9 /L, P -value = .000) and hemoglobin levels (87 g/L vs 107 g/L, P -value = .000) than the RBM group. Both groups exhibited creatinine levels that were within the normal range, although the PCT group exhibited significantly higher values than the RBM group (83.9 μmol/L vs 76.2 μmol/L, P -value = .038). Additionally, the PCT group also demonstrated significantly higher LDH levels (173.5 U/L vs 145.6 U/L, P -value = .000); blood B2M (8.2 mg/L vs 3.91 mg/L, P -value = .000); urine M protein (1.93 mg/24 hours vs 0.19 mg/24 hours, P -value = .012); and BMPC% (65% vs 55%, P -value = .034) than the RBM group. No statistically significant differences in the remaining parameters were observed between the groups (all P  > .05; Table 2 ).

Table 2 Comparison of clinical data. PCT (n = 154) RBM (n = 80) χ 2 / z
P
Age (yr) 62 (54, 67) 59 (52, 66) −1.333 .182 Gender (%) 0.144 .705  Male 102 (66.2) 51 (63.7)  Female 52 (33.8) 29 (36.3) White blood cell (10 9 /L) 4.9 (4.3, 6.2) 4.6 (4, 5.6) −0.741 .458 Platelet (10 9 /L) 149 (113, 191) 192 (170.5, 234) −5.842 .000 Hemoglobin (g/L) 87 (82, 111) 107 (94, 115) −3.703 .000 Uric acid (μmol/L) 472.8 (366.5, 567) 413.45 (347.2, 524.1) −1.737 .082 Urea (mmol/mL) 6.9 (4.6, 8.0) 5.9 (5.3, 7.2) −0.283 .777 Creatinine (μmol/L) 83.9 (71.4, 153.3) 76.2 (63.6, 84.2) −2.073 .038 Blood calcium (mmol/L) 2.4 (2.2, 2.4) 2.3 (2.1, 2.5) −0.775 .438 LDH (U/L) 173.5 (149.4, 242.6) 145.6 (123.9, 194.6) −4.228 .000 Albumin (g/L) 35.3 (29.7, 41.8) 32.4 (30.1, 43.4) −0.216 .829 Blood β2-microglobulin (mg/L) 8.2 (4.55, 13.2) 3.91 (3.24, 6.62) −5.717 .000 Urine M protein (mg/24 h) 1.9 (0, 5.8) 0.2 (0, 2.5) −2.51 .012 BMPC (%) 65 (35, 80) 55 (14.5, 70) −2.124 .034 Blood M protein (g/L) 33.4 (2.5, 56.1) 36.8 (3.2, 53.9) −0.19 .849 Free light chain ratio (κ/λ) 1.8 (0.22, 3.4) 2.8 (0.04, 38.4) −0.593 .553 κ (mg/L) 601 (307, 4200) 566 (145, 3810) −0.839 .401 λ (mg/L) 496 (183, 1920) 193 (113, 3630) −0.499 .618 BMPC = bone marrow plasma-cell percentage, LDH = lactate dehydrogenase, PCT = plasma cell tumor, RBM = red bone marrow.

Comparison of clinical data.

BMPC = bone marrow plasma-cell percentage, LDH = lactate dehydrogenase, PCT = plasma cell tumor, RBM = red bone marrow.

3.3. Comparison of WB-LDCT data

The CT values of lesions in the PCT group were significantly higher than those in the RBM group (70 HU [61, 79] vs 13 HU [6.5, 20.5]; P  = .000). The 95% confidence intervals were (37.4–83.2 HU) and (−12.8 to 38.6 HU), respectively. Significant differences in lesion boundaries were also observed between the 2 groups, with 93.5% of the PCT group exhibiting well-defined boundaries and 90.0% of the RBM group exhibiting poorly defined boundaries ( P -value = .000). The RBM group exhibited a higher proportion of lesions in the proximal half of the humerus/femur than did the PCT group (72.5% vs 49.4%, P  = .001). No statistically significant differences in the remaining parameters were observed between the 2 groups (all P  > .05; Table 3 ).

Table 3 Comparison of WB-LDCT data. PCT (n = 154) RBM (n = 80) χ 2 / z
P
Morphology (%) 97.417 .258  Nodular 63 (40.9) 38 (47.5)  Cast type 91 (59.1) 42 (52.5) Location (%) 0.482 .488  Humerus 73 (47.7) 42 (52.5)  Femur 80 (52.3) 38 (47.5) Extent (%) 11.529 .001  Within the proximal half 76 (49.4) 58 (72.5)  Beyond the proximal half 78 (50.6) 22 (27.5) Area (mm 2 ) 200.5 (78, 545) 259 (95.5, 543) −0.329 .742 CT values (HU) 70 (61, 79) 13 (6.5, 20.5) −12.543 .000 Boundaries (%) 194.494 .000  Poorly defined 10 (6.5) 72 (90.0)  Well-defined 144 (93.5) 8 (10.0) CT = computed tomography, HU = hounsfield unit, PCT = plasma cell tumor, RBM = red bone marrow, WB-LDCT = whole-body low-dose computed tomography.

Comparison of WB-LDCT data.

CT = computed tomography, HU = hounsfield unit, PCT = plasma cell tumor, RBM = red bone marrow, WB-LDCT = whole-body low-dose computed tomography.

3.4. Comparison of diagnostic efficacy

Parameters that exhibited statistically significant differences between the 2 groups were incorporated into the ROC diagnostic model. The AUC of the CT values was 0.926 (sensitivity and specificity of 94.2% and 91.8%, respectively, when using a CT value ≥41.5 HU as the cutoff), indicating a diagnostic accuracy of 93.5%, while that of the lesion boundaries was 0.908, with well-defined boundaries exhibiting a sensitivity of 88.9%, specificity of 93.6%, and diagnostic accuracy of 88.7%. The AUC of the omitted variable (CT value ≥41.5 HU + well-defined boundaries) was 0.914 (sensitivity and specificity, 86.4% and 96.5%, respectively), indicating a diagnostic accuracy of 91.6%. The AUC of all remaining variables was below 0.75, as shown in Table 4 .

Table 4 Comparison of diagnostic efficacy. AUC Sensitivity (%) Specificity (%) PPV (%) NPV (%) Diagnostic accuracy (%) Cutoff CT values (HU) 0.926 (0.905–0.942) 94.2 91.8 89.5 94.6 93.5 41.5 Well-defined 0.908 (0.864–0.931) 88.9 93.6 92.4 86.8 88.7 – CT values ≥ 41.5 HU + well-defined 0.914 (0.892–0.938) 86.4 96.5 95.8 83.4 91.6 – Within the proximal half 0.524 (0.498–0.629) 65.6 52.4 56.2 68.9 49.6 – Platelet (10 9 /L) 0.733 (0.667–0.798) 90.0 65.6 57.6 92.7 73.9 157.5 Creatinine (μmol/L) 0.583 (0.506–0.659) 50.0 81.3 83.7 45.8 60.7 85.05 LDH (U/L) 0.669 (0.597–0.74) 81.8 56.3 78.3 61.6 73.1 146.1 Blood M protein (mg/L) 0.728 (0.661–0.794) 58.4 88.8 90.9 52.6 68.8 7.355 Hemoglobin (g/L) 0.647 (0.574–0.721) 88.8 50.0 48.0 89.5 63.2 86.5 Urine M protein (mg/24 h) 0.599 (0.522–0.676) 49.4 78.8 81.7 44.7 59.4 2.83 BMPC (%) 0.584 (0.506–0.663) 46.8 72.5 76.6 41.4 55.6 67.5 AUC = area under the curve, BMPC = bone marrow plasma-cell percentage, CT = computed tomography, HU = hounsfield unit, LDH = lactate dehydrogenase, NPV = negative predictive value, PPV = positive predictive value.

Comparison of diagnostic efficacy.

AUC = area under the curve, BMPC = bone marrow plasma-cell percentage, CT = computed tomography, HU = hounsfield unit, LDH = lactate dehydrogenase, NPV = negative predictive value, PPV = positive predictive value.

4. Discussion

The current study identified, for the first time, the CT diagnostic parameters (CT value ≥41.5 HU and well-defined boundaries) that facilitated distinguishing between malignant PCTs and benign RBMs. These were excellent diagnostic variables consistent with the diagnostic efficacy of the criteria and may facilitate the imaging diagnosis of MGUS/SMM/MM by using the SLiM (Sixty, light chain, and MRI) standards (M) when WB-MRI is not available or is contraindicated. ROC analysis showed that the sensitivity, specificity, and diagnostic accuracy of a CT value ≥41.5 HU and well-defined boundaries as criteria were 94.2%, 91.8%, 93.5%, and 88.9%, 93.6%, 88.7%, respectively. The sensitivity, specificity, and diagnostic accuracy of the combined variable (CT value ≥41.5 HU + well-defined boundaries) were 86.4%, 96.5%, and 91.6%, respectively. However, given the limitations of the current study, these criteria should not be considered a substitute for WB-MRI. It is important to emphasize that WB-LDCT has a good positive predictive value (94.1%) when osteolytic bone lesions are detected, confirming the diagnosis of MM. However, it has a much weaker negative predictive value (58.8%). Therefore, if there are no bone lesions on WB-LDCT, then a follow-up investigation with WB-MRI (or PET/CT if WB-MRI is not available) should be performed to exclude lesions and confirm the diagnosis of smoldering MM or to find lesions that are undetectable with WB-LDCT (2 or more lesions ≥ 5 mm at MRI) and upgrade the diagnosis to MM. [ 4 ] Furthermore, 93.5% of lesions in the PCT group exhibited well-defined boundaries, and 90.0% of those in the RBM group exhibited poorly defined boundaries, potentially due to the greater contrast in CT values between PCTs and RBMs adjacent to the yellow bone marrow.

Matsue et al [ 10 ] found that the CT value of yellow bone marrow was −68.4 HU (range: −92.8 to −44.3 HU), and other researchers proposed a diagnostic cutoff value of CT ≥−30 HU to distinguish lesions from yellow bone marrow. [ 8 , 13 , 14 ] Horger et al reported that the mean CT value of lesions within the medullary cavities ranged between 52 and 65 HU, and these were lower than the CT values observed in the PCT group (i.e., 70 HU, range: 61–79 HU) and higher than those observed in the RBM group (i.e., 13 HU, range: 6.5–20.5 HU) in the current study. [ 8 – 10 , 13 , 14 ] The CT values of lesions in the PCT group were significantly higher than those in the RBM group (70 HU [61, 79] vs 13 HU [6.5, 20.5]; P  = .000). The 95% confidence intervals are (37.4–83.2 HU) and (−12.8 to 38.6 HU), respectively, consistent with the pathological differences in lesion composition. The 95% confidence interval can better reflect the difference between the means of the 2 groups. [ 17 ] Specifically, the RBMs comprised 60% cellular components and 40% fat cells, whereas the PCTs consisted of dense abnormal plasma cell clusters and were almost devoid of fat. [ 6 ]

Medullary lesions of AS were first detected using WB-LDCT in 2008 by Horger et al [ 8 ] and were later classified into focal, diffuse, and mixed types by other researchers. [ 9 , 10 , 13 , 14 ]

The current study included individual lesions ≥10 mm in size and further classified the mixed lesions into nodular and cast types. No statistically significant differences in the morphology of the lesions were observed between the PCT and RBM groups (nodular type: 40.9% vs 47.5%; cast type: 59.1% vs 52.5%). Furthermore, no significant differences in lesion location (humerus: 47.7% vs 52.5%; femur: 52.3% vs 47.5%) were observed between the 2 groups. This is in contrast to the findings of Koutoulidis et al, [ 9 ] who observed involvement of the humerus and femur in 45/127 and 82/127 cases, respectively. These differences in outcomes may be attributed to the inclusion of patients with MM or smoldering MM. The RBM group exhibited a higher proportion of lesions in the proximal half of the humerus/femur than the PCT group, in line with the pattern of red-to-yellow bone marrow conversion. In contrast, the distribution of PCTs was similar between the proximal and distal ends (49.4% and 50.6%, respectively).

The PCT group exhibited significantly lower platelet counts and hemoglobin levels than the RBM group, indicating reduced hematopoiesis, a higher degree of bone marrow infiltration, and more severe damage in the PCT group. The creatinine levels in the PCT group were also significantly higher than those in the RBM group (although both were still within the normal range), suggesting severe renal dysfunction in the former, consistent with the results reported by Matsue et al. [ 10 ] Additionally, the PCT group exhibited significantly higher LDH, blood B2M, urine M protein, and BMPC% compared with the RBM group, suggesting that the overall tumor burden and the degree of bone infiltration and destruction were higher in the former group than in the latter. Horger et al [ 8 ] found that patients with medullary lesions exhibited higher levels of blood B2M than those without such lesions (4.3 mg/dL vs 2.4 mg/dL, P  = .0015), which was consistent with the findings of the current study. Although the above-mentioned clinical parameters can reflect a higher tumor burden and severe end-organ damage in the PCT group, their AUCs were <0.75, indicating an insignificant diagnostic value. The absence of any statistically significant differences in the remaining clinical parameters between the 2 groups suggested that they were not associated with the nature of the lesions.

This study has several limitations. First, the pathological findings were not included in the diagnostic criteria for PCTs. Although the use of WB-MRI with multiple parameters and follow-up examinations is a well-established technique for differentiating between PCTs and RBM, it is still inferior to pathological diagnosis, which is considered the gold standard method in terms of accuracy. Second, the proportion of MGUS and SMM patients included in this study was small (14/234 and 35/234, respectively), with mainly MM patients (185/234), which cannot represent all patients with plasma cell diseases despite the consistency of the lesions. A large sample population of patients with MGUS/SMM will be prospectively included in the later stage. Third, to minimize measurement errors, the current study included only nodular and cast lesions with a diameter of ≥10 mm. Therefore, these findings may not be applicable to other types of lesions (e.g., mixed lesions). Fourth, the CT value, which is the most crucial diagnostic parameter, was measured manually using a polygonal tool and was subject to volumetric effects, which may have resulted in errors. The boundary parameter (i.e., well-defined and poorly defined) was subjectively evaluated as a binary variable by physicians, which may have introduced bias due to potential inconsistencies between practitioners.

In conclusion, a CT value ≥41.5 HU and well-defined boundaries were both excellent diagnostic variables for PCTs, and the combined variable (CT value ≥41.5 HU + well-defined boundaries) had higher diagnostic specificity.

Author contributions

Conceptualization: Weiwei Lu, Wenyang Huang, Ying Li.

Data curation: Kemin Zhang, Bing Shi, Jinshan Wang, Xiaoli Wang, Yunqing Zhao, Xiaoxue Wang, Haijing Xian, Wentao Men, Mingyue Liu, Xiaodong Ji.

Formal analysis: Kemin Zhang, Shanyan Wang, Bing Shi, Yunqing Zhao, Xiaoxue Wang, Haijing Xian, Wentao Men, Mingyue Liu.

Funding acquisition: Wenyang Huang.

Investigation: Kemin Zhang, Shanyan Wang, Jinshan Wang, Xiaoli Wang, Shuaishuai Zhang.

Methodology: Weiwei Lu, Hong Dang, Xiaodong Ji, Ying Li.

Project administration: Wentao Men, Mingyue Liu.

Resources: Yunqing Zhao, Haijing Xian, Shuaishuai Zhang.

Software: Shanyan Wang, Weiwei Lu, Xiaoxue Wang, Xiaodong Ji.

Validation: Hong Dang.

Visualization: Bing Shi, Weiwei Lu.

Writing – original draft: Kemin Zhang.

Writing – review & editing: Wenyang Huang.

Abbreviations: ADC apparent diffusion coefficient AS appendicular skeleton AUC area under the curve B2M β2-microglobulin BMPC% bone marrow plasma-cell percentage CT computed tomography DWI diffusion-weighted imaging HU hounsfield unit IP in-phase LDH lactate dehydrogenase MGUS monoclonal gammopathy of undetermined significance MM multiple myeloma MPR multiplanar reconstruction OP out-of-phase PCT plasma cell tumor PET positron emission tomography RBM red bone marrow ROC receiver operating characteristic SMM smoldering multiple myeloma SR suppression ratio TE echo time TR repetition time WB-LDCT whole-body low-dose computed tomography WB-MRI whole-body magnetic resonance imaging

Abbreviations:

apparent diffusion coefficient

appendicular skeleton

area under the curve

β2-microglobulin

bone marrow plasma-cell percentage

computed tomography

diffusion-weighted imaging

hounsfield unit

in-phase

lactate dehydrogenase

monoclonal gammopathy of undetermined significance

multiple myeloma

multiplanar reconstruction

out-of-phase

plasma cell tumor

positron emission tomography

red bone marrow

receiver operating characteristic

smoldering multiple myeloma

suppression ratio

echo time

repetition time

whole-body low-dose computed tomography

whole-body magnetic resonance imaging

This study was supported by the National Natural Science Foundation of China (82070192) and the Clinical and Translational Medicine Research Project of the Medical and Health Technology Innovation Project, Chinese Academy of Medical Sciences (2021-I2M-C&T-B-079).

The authors have no conflicts of interest to declare.

The datasets generated and/or analyzed during the current study are available from the corresponding author on reasonable request.

How to cite this article: Zhang K, Wang S, Shi B, Wang J, Wang X, Lu W, Zhao Y, Wang X, Dang H, Xian H, Men W, Liu M, Zhang S, Ji X, Huang W, Li Y. Using WB-LDCT to differentiate between benign and malignant medullary lesions in the appendicular skeleton of patients with plasma cell diseases. Medicine 2026;105:37(e50461).

The funding sources had no role in study design, data collection and analysis, manuscript preparation, or the decision to publish.

Contributor Information

Kemin Zhang, Email: zhangshuaishuai@ihcams.ac.cn.

Shanyan Wang, Email: wangxiaoxue@ihcams.ac.cn.

Bing Shi, Email: tsshibing79@126.com.

Jinshan Wang, Email: wangxiaoxue@ihcams.ac.cn.

Xiaoli Wang, Email: wangxiaoxue@ihcams.ac.cn.

Weiwei Lu, Email: 474933203@qq.com.

Yunqing Zhao, Email: zhaoyunqing@ihcams.ac.cn.

Xiaoxue Wang, Email: wangxiaoxue@ihcams.ac.cn.

Hong Dang, Email: 415723972@qq.com.

Haijing Xian, Email: 1297033404@qq.com.

Wentao Men, Email: menwentao1995@vip.qq.com.

Mingyue Liu, Email: 1179268086@qq.com.

Shuaishuai Zhang, Email: zhangshuaishuai@ihcams.ac.cn.

Xiaodong Ji, Email: lz_east@126.com.

Wenyang Huang, Email: huangwenyang@ihcams.ac.cn.

References