Circulating sRANKL, Periostin, and Osteopontin as Biomarkers for the Assessment of Activated Osteoclastogenesis in Myeloma Related Bone Disease

Simple Summary Multiple myeloma (MM) accounts for about 10% of all hematologic malignancies and is usually accompanied by severe bone disease. We analyzed the circulating levels of sRANKL, periostin, and osteopontin as osteoclast activators in newly diagnosed MM (NDMM) patients at diagnosis and in the course of treatment. Given the significantly higher levels of tested proteins at diagnosis, their correlations with the stage and grade of the disease, as well as their gradual decrease in the course of treatment, we suggest the applicability of these proteins as potential biomarkers for myeloma related bone disease (MBD) monitoring. Abstract The hallmark of multiple myeloma is myeloma related bone disease. Interactions between myeloma plasma cells (MPCs), stromal cells, and the bone marrow (BM) microenvironment play a critical role in the pathogenesis of MBD. Bone remodeling is severely dysregulated with the prevalence of osteoclast activity. We aimed to assess circulating levels of sRANKL, periostin, and osteopontin as osteoclast activators in NDMM patients at diagnosis and in the course of treatment, correlations with clinical and laboratory data, and to evaluate their potential as additional biomarkers for the assessment of MBD. The current study involved 74 subjects (41 NDMM patients, 33 controls). MBD was assessed by whole-body low-dose computed tomography. sRANKL, periostin, and osteopontin were assayed by commercial ELISA kits. At diagnosis, all tested parameters were significantly higher in NDMM patients compared to the controls (p < 0.0001), correlating with disease stage, MBD grade, and BM infiltration by MPCs. During therapy, the serum levels of all tested proteins decrease, most prominently after autologous stem cell transplantation (p < 0.0001). A significant reduction was established in patients achieving complete and very-good partial response compared to all others (p < 0.05). In conclusion, sRANKL, periostin, and osteopontin reflect MBD severity and could be promising markers for MBD monitoring and the effect of myeloma treatment.


Introduction
Multiple myeloma (MM) is a plasma cell disorder with increasing incidence worldwide in recent decades [1].It accounts for about 10% of all hematologic malignancies [2].The uncontrolled proliferation of abnormal monoclonal plasma cells (MPCs) in bone marrow (BM) impels the development of myeloma related bone disease (MBD).At diagnosis, up to 80% of patients present with osteolytic bone lesions [3] and are at increased risk of skeletal-related events (SRE) such as pathological fractures and spinal cord compression.Bone complications lead to a significant reduction of quality of life and place a substantial additional burden on patients and healthcare systems [4].Complex interactions between MPCs, stromal cells, and the BM microenvironment play a critical role in the pathogenesis of MBD [5,6].Many signaling pathways including Notch signaling, Wnt/β-Catenin signaling, RANKL/RANK/OPG (receptor activator of nuclear factor kappa-B ligand/receptor activator of NF-κB, osteoprotegerin) axis are activated [7,8].Bone remodeling is severely dysregulated, manifested by increased osteoclast activity and suppressed osteoblast activity resulting in osteolytic lesions and bone destruction [9].RANKL/RANK signaling is vital for osteoclastogenesis.As a membrane protein, RANKL is highly expressed in osteoblasts, osteocytes, and activated T lymphocytes [10].The soluble form of the protein (sRANKL) is generated by either proteolytic processing by matrix metalloproteases or alternative mRNA splicing [11].RANKL binds to RANK on the surface of osteoclasts and osteoclast precursors, leading to their differentiation and activation [12,13].In the 5T2MM mouse model of multiple myeloma, MPCs overexpress RANKL, which directly promotes the formation of osteoclasts and osteolytic lesions [14].Terpos et al. first reported elevated serum levels of sRANKL in MM patients and suggested that RANKL/OPG system is essential for the development of MBD [15].The early involvement of RANKL signaling was observed in patients with monoclonal gammopathy of undetermined significance (MGUS), where circulating sRANKL levels were higher than in controls but still lower than in patients with symptomatic MM [16].Serum RANKL was significantly increased in advanced clinical stages and high grade MBD of MM patients [17,18].Therefore, inhibition of the RANK/RANKL pathway is an important target for the prevention of SRE in MM.Denosumab, a fully human monoclonal anti-RANKL antibody, has been in use for more than ten years in clinical practice [19].
Another protein associated with bone metabolism is the extracellular matrix (ECM) protein periostin, which is produced primarily by stromal mesenchymal cells and is mainly expressed in the periosteum [20,21].In addition to its structural role in the bone matrix, under physiological conditions, periostin stimulates osteoblastic function and bone formation activating the Wnt/β-catenin signaling pathway [22].Under pathological conditions, high serum periostin is associated with an increased risk of fractures in postmenopausal women [23,24], and in ankylosing spondylitis [25].Many studies have reported elevated periostin expression in primary tumors and metastatic lesions, e.g., colorectal cancer and liver metastatic lesions [26][27][28][29][30][31].There are very scarce data concerning periostin implication in MM.Terpos et al. first demonstrated high periostin in the supernatant of six myeloma cell lines.Beside the experimental data, the authors investigated periostin in clinical settings and found the progressive elevation of its levels as the disease evolves from MGUS, SMM (smoldering multiple myeloma), and symptomatic MM [32].Revealing the role of periostin in the pathogenesis of MBD, target therapy with anti-periostin antibodies might be a promising new treatment strategy.
Another abundant non-collagenous ECM protein that is an important component of the BM microenvironment is osteopontin (OPN) [33].It is expressed by a wide variety of cell types such as bone cells, neurons, epithelial cells, pericytes, fibroblasts, hepatocytes, tubular cells, vascular smooth muscle cells, and immune cells [34].It is involved in a number of physiological and pathological processes, including bone remodeling, angiogenesis, immune-regulation, inflammation, wound healing, atherosclerosis, and tumor progression [33,[35][36][37][38].In models with human MM cell lines, Abe et al. demonstrated a cell-cell contact-mediated mechanism of OC induced MPC growth and survival, partially dependent on IL-6 and OPN.A vicious cycle is supposed to be formed, in which the production of OPN along with IL-6 from OCs recruited around MPCs contribute to extensive bone destruction and MPC expansion [39].Various studies have found significantly elevated plasma OPN levels in symptomatic MM patients compared to MGUS patients, patients in remission and healthy controls [40][41][42].OCs-derived osteopontin and VEGF from MPCs cooperatively enhance angiogenesis and also induce osteoclastogenic activity by vascular endothelial cells [43].
The aim of the present study was to explore circulating levels of sRANKL, periostin, and osteopontin in newly diagnosed MM (NDMM) patients at diagnosis and in the course of their treatment, including autologous stem cell transplantation (ASCT), and to assess possible correlations with laboratory data, disease stage, and MBD.

Study Population and Study Design
A total of 41 NDMM patients were enrolled in this single-center non-interventional prospective study, carried out between June 2021 and December 2022.Selected patients met the following inclusion criteria: 18 to 80 years of age; and diagnosed according to the Revised International Myeloma Working Group criteria [44].Patients with autoimmune or other malignant diseases and treated with bone affecting drugs during the last six months were excluded from the study.The disease stage was determined according to the International Staging System, ISS [45].The study design was explained in detail in [46].In brief, whole-body low-dose computed tomography (WBLDCT) imaging was used for the assessment of MBD, defined by osteolytic lesions, pathological fractures, vertebral collapse, and soft-tissue involvement.All patients were treated using the standard VCD chemotherapeutic protocol.In addition, patients with a glomerular filtration rate of more than 30 mL/min/1.73m 2 received biphosphonates.International Myeloma Working Group (IMWG) consensus criteria for response were used to evaluate response to treatment [47].After they completed four cycles of VCD treatment, the patients were divided into two groups: patients below 65 years and with a complete response (CR) and very good partial response (VGPR) proceeded to ASCT; patients with stable disease (SD) or partial response (PR) continued with another four cycles of VCD.Patients with progressive disease in the course of treatment dropped out from the study.Four time points for the evaluation of the variations in the tested biochemical markers were selected: before treatment (time point T0); after four cycles of VCD treatment (time point T1); after eight cycles of VCD treatment (time point T2); three months after ASCT (time point TA).
Thirty-three healthy age-and gender-matched control subjects without any chronic inflammatory, autoimmune, or neoplastic disease or bisphosphonate and other boneaffecting drug medication six months before starting the survey were included in the study.

Methods
Serum sRANKL, periostin, and osteopontin were assayed by commercial ELISA (Sunredbio, Shanghai, China), strictly following the manufacturer's instructions.Hematological and biochemical parameters as well as morphological evaluations of bone marrow aspirate smears were routinely monitored.

Statistical Analysis
Results were presented as the mean ± standard deviation (SD), median, and interquartile range (IQR: 25-75-th percentile), or number (n) and percentage (%), as appropriate.Descriptive statistics, the unpaired Student's t-test, and the Mann-Whitney U-test for comparison of Gausian or non-Gausian distributed interval data were used.Chi-square analysis was used to evaluate the frequency distribution according to gender.The relationship between continuous variables was evaluated by Spearman's correlation analysis, and if a significant relation was found, a linear regression analysis was applied to test the associations of sRANKL, periostin, and osteopontin with routine laboratory parameters.Receiver operating characteristic (ROC) analysis was performed for evaluation the predictive power of serum sRANKL, periostin, and osteopontin.All statistical analyses were two-tailed and statistical significance was considered at p < 0.05.Data analysis was performed using GraphPad Prism v. 8.0.1,San Diego, CA, USA.
Receiver operating characteristic (ROC) analysis was performed for evaluation the predictive power of serum sRANKL, periostin, and osteopontin.All statistical analyses were two-tailed and statistical significance was considered at p < 0.05.Data analysis was performed using GraphPad Prism v. 8.0.1,San Diego, CA, USA.

Results
The baseline demographic and clinical characteristics, as well some results from the routine laboratory testing of MM patients enrolled in the study are provided in detail in [46] and in Supplementary Table S1.

Variation of sRANKL, Periostin, and Osteopontin Serum Levels in the Course of Treatment and According to the Response to Therapy
The medians and corresponding IQR of sRANKL, periostin, and osteopontin serum levels at different time points in the course of treatment are given in Supplementary Table S3.Serum levels for sRANKL, periostin, and osteopontin gradually decreased in the course of treatment.A significant reduction in periostin was already observed at time point T1, while sRANKL and osteopontin were non-significantly lower than the baseline values (T0).For patients, not eligible for transplantation, the periostin levels continued to   S3.
Serum levels for sRANKL, periostin, and osteopontin gradually decreased in the course of treatment.A significant reduction in periostin was already observed at time point T1, while sRANKL and osteopontin were non-significantly lower than the baseline values (T0).For patients, not eligible for transplantation, the periostin levels continued to decrease and remained significantly lower at time point T2 compared to the T0 values (p = 0.0024), while the sRANKL and osteopontin changes continued to not show any significant reduction.Patients who underwent ASCT, showed the lowest values for all measured proteins at TA, significantly different from those in T0 and T1 (Figure 6).Examining the tested parameter changes in the course of treatment in relation to the control values, we found significantly higher serum periostin at T1 (p = 0.0005), while at T2 and TA it almost reached the control levels (p = 0.1899 and p > 0.9999, respectively).The same phenomenon was observed for sRANKL and osteopontin only at the TA time point (p = 0.5614 and p = 0.0996, respectively).
The response to therapy at all treatment points is summarized in Table 1.Patients with a CR and VGPR revealed significantly lower levels for the tested parameters vs. baseline levels (T0).In addition, periostin and osteopontin almost reached the control values, while sRANKL remained significantly higher (p = 0.0036).On the other hand, sRANKL and osteopontin remained unchanged compared to T0 in PR and SD patients, while periostin was significantly decreased.Regarding the controls, the patients Examining the tested parameter changes in the course of treatment in relation to the control values, we found significantly higher serum periostin at T1 (p = 0.0005), while at T2 and TA it almost reached the control levels (p = 0.1899 and p > 0.9999, respectively).The same phenomenon was observed for sRANKL and osteopontin only at the TA time point (p = 0.5614 and p = 0.0996, respectively).
The response to therapy at all treatment points is summarized in Table 1.Patients with a CR and VGPR revealed significantly lower levels for the tested parameters vs. baseline levels (T0).In addition, periostin and osteopontin almost reached the control values, while sRANKL remained significantly higher (p = 0.0036).On the other hand, sRANKL and osteopontin remained unchanged compared to T0 in PR and SD patients, while periostin was significantly decreased.Regarding the controls, the patients from this group revealed significantly higher values for all measured proteins (Figure 7).

Predicting Power of sRANKL, Periostin, and Osteopontin
ROC analysis was performed to evaluate the diagnostic utility of the tested parameters for MBD assessment in NDMM patients (Table 3 and Figure 8).The area under the ROC curve for each of the tested parameters is above 0.85 (p < 0.0001), revealing a relatively high distinguishing capacity between positive and negative cases of MBD in NDMM patients.All tested proteins revealed high diagnostic sensitivity and specificity at the most appropriate cut-off value.
Multiple linear regression analysis was used to estimate the relationship between each one of tested parameters as a dependent variable and several independent variables

Predicting Power of sRANKL, Periostin, and Osteopontin
ROC analysis was performed to evaluate the diagnostic utility of the tested parameters for MBD assessment in NDMM patients (Table 3 and Figure 8).

Predicting Power of sRANKL, Periostin, and Osteopontin
ROC analysis was performed to evaluate the diagnostic utility of the tested parameters for MBD assessment in NDMM patients (Table 3 and Figure 8).The area under the ROC curve for each of the tested parameters is above 0.85 (p < 0.0001), revealing a relatively high distinguishing capacity between positive and negative cases of MBD in NDMM patients.All tested proteins revealed high diagnostic sensitivity and specificity at the most appropriate cut-off value.
Multiple linear regression analysis was used to estimate the relationship between each one of tested parameters as a dependent variable and several independent variables with which a significant correlation was established.For sRANKL linear regression was The area under the ROC curve for each of the tested parameters is above 0.85 (p < 0.0001), revealing a relatively high distinguishing capacity between positive and negative cases of MBD in NDMM patients.All tested proteins revealed high diagnostic sensitivity and specificity at the most appropriate cut-off value.

Discussion
The pathogenesis of MBD is multifactorial.Interactions of MPCs with various cell types such as lymphocytes, endothelial and stromal cells, as well as components of the extracellular bone marrow microenvironment play a crucial role in the development of MBD.In the last two decades, several cytokines and ECM proteins have been evaluated as biomarkers for the assessment of MBD.We analyzed the serum levels of sRANKL, periostin, and osteopontin in NDMM patients depending on the clinical stage and grade of MBD and their variations in the course of therapy to evaluate their potential as additional biomarkers for the assessment of MBD.

Baseline Serum Levels of sRANKL, Periostin, and Osteopontin in Newly Diagnosed Multiple Myeloma Patients
The role of sRANKL as a potent osteoclast activating factor is well established in animal and in vitro models [14,[50][51][52].mRNA and soluble RANKL have been observed in samples from MM patients, but not in chronic lymphocytic leukemia, indicating a unique requirement for RANKL in multiple myeloma [53].In our study the sRANKL serum levels in NDMM patients at baseline were significantly higher than those in controls.In agreement with our results are the findings of most previous studies, which have demonstrated significantly higher values for sRANKL in NDMM patients vs. control subjects [15][16][17]54].A few reports did not find a statistical difference in the serum levels of sRANKL in NDMM patients compared to controls, and no differences were observed with respect to disease stage and bone lesions [55,56].The inconsistency of the abovementioned results could be explained by the different ELISA kits used for sRANKL determination.It is not surprising that RANKL, as a potent osteoclast activator, increases with osteoclast bone lesions as well as with the disease stage.In our study lower RANKL levels were associated with a small number of osteolytic lesions (<3, G1), whereas higher RANKL was related to advanced bone disease (G2).Similar relationships were observed for circulating RANKL and disease stage, increased in ISS-2 and ISS-3, and lower in ISS-1.Our findings agree with other studies reporting a significant increase of serum RANKL with advanced disease stage, grade of bone disease, and factors of disease activity such as B2MG and BMI [15][16][17][18]54,57].
The role of periostin in the pathogenesis of MM is poorly understood.In an experimental study it was suggested that fibroblast-like cells from the BM microenvironment are implicated in ECM remodeling and lead to the upregulation of various matrix proteins, including periostin, thus creating a favorable environment for MPC development.The upregulation takes place already at the level of MGUS and is more pronounced in MM [58].Data from clinical studies regarding periostin are really scarce.Terpos et al. were the first to show significantly increased serum levels of periostin in NDMM and relapsed patients compared not only to healthy subjects, but also to MGUS and SMM patients.The authors found that serum periostin strongly correlated with B2MG and ISS-stages.The established correlations of periostin (positive with CTX, a bone resorption marker, and negative with bone alkaline phosphatase, a bone formation marker) suggest the role of periostin in the development of MBD.Moreover, patients with pathological fractures had markedly elevated periostin both in serum and BM plasma in comparison with all others.Patients with osteolytic lesions had significantly higher periostin in BM, but not in serum compared to those without osteolytic lesions [32].Consistent with these data are our results which demonstrate a significant increase in serum periostin levels in NDMM patients vs. controls.We also established that periostin levels progressively increased both with the disease stages and the severity of bone involvement.In contrast with the finding of Terpos et al. we found significant differences in periostin serum levels in the G2 group vs. G1 group.The discordance may be due to the fact that patients with pathological fractures were also included in the G2 group.Also findings a moderate to strong positive correlation of serum periostin with B2MG, BMI, creatinine, and TP, as well as a negative correlation with albumin and Hb, all of which are associated with disease progression, we suggest the important role of periostin for MPC invasion in the BM and for MBD.
OPN, as the most abundant non-collagenous protein in the bone ECM, is an integral part of the BM microenvironment, where it regulates matrix interactions and cell adhesion [59].For bone resorption, OPN-mediated attachment of OCs to the bone surface represents a crucial step.The role of OPN in MM is better studied than that of periostin, but different results have been reported.In most studies, the baseline OPN levels in symptomatic patients were significantly higher than those of controls [40][41][42]60,61].We have also obtained similar results.On the contrary, Kang et al. did not find a significant difference in OPN levels between MM patients and healthy controls [62].Moreover, in a recent study, even significantly lower OPN levels were found in stage I MM patients [63].This discrepancy is probably due to the small patient sample, poorly defined patients' disease stages, and the ELISA method used for osteopontin determination.Our data revealing statistically higher levels in ISS-III patients compared to ISS-I patients, as well in the G2 vs. G1 group, confirm the results of several studies that OPN levels reflect the progression of MM and MBD well [42,61,64].In addition, we proved that patients with BMI ≥ 60% MPCs had significantly higher levels of OPN than those with BMI < 60% MPCs.OPN is produced not only by osteoclasts, but also by MPCs and in turn elevated OPN stimulates MPCs proliferation and exerts a proangiogenic effect on BM blood vessels [39,65,66], thus forming a vicious cycle of the self-maintenance and progression of the tumor process.Sfiridaki et al. reported a significant correlation between serum OPN and disease infiltration, as well with serum vascular endothelial growth factor (VEGF) and BM microvessel density (MVD) as markers of BM neovascularization [64].In an Interesting in vitro study Robbiani et al. showed that high OPN expression correlates inversely with bone disease, explained by presence of maf translocations [67].In the present study, the association of OPN with tumor burden was also proven by a strong positive correlation with B2MG and a weak negative correlation with albumin and hemoglobin.Similar results were reported by other researchers [41,60,61,64].

Variation of sRANKL, Periostin, and Osteopontin Serum Levels in the Course of Treatment and According to the Response to Therapy
Bortezomib-based regimens are most frequently used for MM [68].In addition to an antimyeloma effect, Bortezomib (Velcade) induces mesenchyme stem cells to preferentially undergo osteoblastic differentiation and suppresses osteoclasts, ultimately leading to bone formation [69,70].
In the first clinical study concerning the effect of Bortezomib administration on bone remodeling markers, Terpos et al. have shown significantly reduced serum levels of sRANKL, DKK-1, CTX, and TRACP-5b after four cycles of therapy, irrespective of treatment response, thus leading to the normalization of bone remodeling in relapsed myeloma patients [71].In several other clinical studies bortezomib-based therapies also have led to a significant reduction of serum RANKL [64,69].Our results are consistent with the abovementioned findings showing a gradual decrease of sRANKL in the course of treatment.Moreover, we found significantly lower levels of this cytokine in NDMM patients with a CR and VGPR vs. those with a PR and SD.
In our study, circulating periostin showed a strong association with the course of treatment.In the different time points of chemotherapy or after transplantation, its serum levels were significantly lower than those at diagnosis.Moreover, the patients with a CR and VGPR revealed significantly lower periostin levels compared to those with a PR and in stable disease.To our knowledge, we are the first to demonstrate that circulating periostin reflects the effectiveness of therapy.
In comparison with sRANKL, considerably less data are available in the literature regarding the effect of therapy on serum OPN levels.Sfiridaki et al. have shown significantly lower OPN after chemotherapy [64].Minarik et al. found not only lower post-chemotherapeutic OPN levels, but also significantly lower levels in patients with a deep response in comparison with initial OPN levels [42].In our study, we examined OPN level variations in the course of treatment and found a gradual decrease with the lowest levels achieved 3 months after ASCT.
According to the ROC analysis, the three biomarkers show similar diagnostic reliability (AUC approximately 0.9, p < 0.0001).Depending on the ISS-stage, serum periostin and osteopontin differentiate the intermediate ISS-II, which was not observed for sRANKL (Figure 2).According to treatment response, periostin most adequately reflects the treatment effect.In patients with a CR + VGPR, periostin serum levels decrease and almost reach control levels.In patients with a PR + SD, they remain elevated vs. the controls and a CR + VGPR but are lower than baseline levels.Based on these findings, we could hypothesize that serum periostin might be the most sensitive indicator among the tested parameters.Further studies involving larger cohorts should be conducted to confirm this hypothesis.
As these ECM proteins have been very scarcely studied in clinical settings, we consider the monitoring of OPN, PON, and RANKL serum levels in NDMM patients and the effect of therapy on their serum levels to constitute the strengths of our study.There are also some limitations in our study.One limitation is the small number of studied patients.Another is that we did not examine the changes in osteoprotegerin (OPG) levels in NDMM patients and its interrelations with the proteins that we tested, especially with sRANKL.

Conclusions
MBD has a significant impact on patient quality of life.The discovery and validation of sensitive serum bone markers would aid in the prediction of skeletal complications and their early treatment, improving patient outcomes and reducing the burden on healthcare systems.The present study demonstrates activated osteoclastogenesis, evident by elevated levels of bone resorption cytokines sRANKL, periostin, and osteopontin in NDMM.Moreover, their increased levels strongly correlate with the MM stage and the grade of MBD.Studying the dynamics of these proteins, we demonstrated a significant decrease during treatment, more pronounced after ASCT.In the context of these findings, sRANKL, periostin, and osteopontin could be promising markers for monitoring MBD in MM and the effect of treatment.

Supplementary Materials:
The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/cancers15235562/s1.Table S1: Patients' demographic and clinical characteristics at baseline.Table S2: Correlations of sRANKL, periostin and osteopontin with some laboratory parameters; Table S3: Serum values of sRANKL, Periostin and Osteopontin, measured at different time points.

Figure 1 .
Figure 1.Serum levels of sRANKL, periostin, and osteopontin in controls and patients at baseline.Data are presented as medians and minimal and maximal values.Statistical significance was indicated at p < 0.05.

Figure 1 .
Figure 1.Serum levels of sRANKL, periostin, and osteopontin in controls and patients at baseline.Data are presented as medians and minimal and maximal values.Statistical significance was indicated at p < 0.05.

Cancers 2023 , 16 Figure 2 .
Figure 2. Comparison of sRANKL, periostin, and osteopontin serum levels in NDMM patients at baseline (T0) and stratified according to the ISS staging system.Data are presented as medians and minimal and maximal values.Statistical significance was indicated at p < 0.05.

Figure 2 .
Figure 2. Comparison of sRANKL, periostin, and osteopontin serum levels in NDMM patients at baseline (T0) and stratified according to the ISS staging system.Data are presented as medians and minimal and maximal values.Statistical significance was indicated at p < 0.05.

Figure 2 .
Figure 2. Comparison of sRANKL, periostin, and osteopontin serum levels in NDMM patients at baseline (T0) and stratified according to the ISS staging system.Data are presented as medians and minimal and maximal values.Statistical significance was indicated at p < 0.05.

Figure 3 .
Figure 3. Serum levels of sRANKL, periostin and osteopontin in NDMM patients, stratified by the number of bone lesions.Data are presented as medians and minimal and maximal values.Statistical significance was indicated at p < 0.05.

Figure 3 .
Figure 3. Serum levels of sRANKL, periostin and osteopontin in NDMM patients, stratified by the number of bone lesions.Data are presented as medians and minimal and maximal values.Statistical significance was indicated at p < 0.05.

Figure 4 .
Figure 4. Serum levels of sRANKL, periostin, and osteopontin in NDMM patients, stratified according to BMI by MPCs.Data are presented as medians and minimal and maximal values.Statistical significance was indicated at p < 0.05.

Figure 4 .
Figure 4. Serum levels of sRANKL, periostin, and osteopontin in NDMM patients, stratified according to BMI by MPCs.Data are presented as medians and minimal and maximal values.Statistical significance was indicated at p < 0.05.

3. 2 .
Variation of sRANKL, Periostin, and Osteopontin Serum Levels in the Course of Treatment and According to the Response to Therapy The medians and corresponding IQR of sRANKL, periostin, and osteopontin serum levels at different time points in the course of treatment are given in Supplementary Table

Cancers 2023 , 16 Figure 7 .
Figure 7. sRANKL, periostin, and osteopontin according to the treatment response.Data are presented as medians and minimal and maximal values.Statistical significance was indicated at p < 0.05.
operator characteristics; AUC-area under the ROC curve; CI-confidential interval.

Figure 8 .
Figure 8. Diagnostic performance of sRANKL, periostin, and osteopontin assessed by ROC curve analysis.ROC curve-receiver operating characteristic curve, AUC-area under the curve.The green dotted line represents the settings for finding the optimal diagnostic cut-off values of the tested parameters.Statistical significance was indicated at p < 0.05.

Figure 7 .
Figure 7. sRANKL, periostin, and osteopontin according to the treatment response.Data are presented as medians and minimal and maximal values.Statistical significance was indicated at p < 0.05.

Figure 7 .
Figure 7. sRANKL, periostin, and osteopontin according to the treatment response.Data are presented as medians and minimal and maximal values.Statistical significance was indicated at p < 0.05.
operator characteristics; AUC-area under the ROC curve; CI-confidential interval.

Figure 8 .
Figure 8. Diagnostic performance of sRANKL, periostin, and osteopontin assessed by ROC curve analysis.ROC curve-receiver operating characteristic curve, AUC-area under the curve.The green dotted line represents the settings for finding the optimal diagnostic cut-off values of the tested parameters.Statistical significance was indicated at p < 0.05.

Figure 8 .
Figure 8. Diagnostic performance of sRANKL, periostin, and osteopontin assessed by ROC curve analysis.ROC curve-receiver operating characteristic curve, AUC-area under the curve.The green dotted line represents the settings for finding the optimal diagnostic cut-off values of the tested parameters.Statistical significance was indicated at p < 0.05.

Table 1 .
Frequency distribution according to treatment response.

Table 2 .
Serum levels of sRANKL, periostin, and osteopontin according to treatment response.

Table 1 .
Frequency distribution according to treatment response.

Table 2 .
Serum levels of sRANKL, periostin, and osteopontin according to treatment response.

Table 3 .
Diagnostic utility of the tested parameters evaluated by ROC curve analysis.

Table 3 .
Diagnostic utility of the tested parameters evaluated by ROC curve analysis.

Table 3 .
Diagnostic utility of the tested parameters evaluated by ROC curve analysis.