Evaluation of Bone Mineral Density and Fat-Lean Distribution in Patients With Multiple Myeloma in Sustained Remission


  • The authors have no conflict of interest


To study the usefulness of bone mineral density (BMD) in the follow-up of myeloma (MM) patients, BMD was evaluated in 44 MM patients in sustained remission for at least 2 years (35.4 ± 10.5 months) after high-dose or conventional chemotherapy in a retrospective study. Patients never received bisphosphonates before or during the follow-up. Patients underwent lumbar spine (LS) BMD and a whole body (WB) BMD testing before therapy and at least once in the remission period. At baseline, mean LS BMD was 0.863 ± 0.026 g/cm2, mean lumbar Z-score was −1.45 SD. LS BMD significantly increased from baseline by 5 ± 1.8%, 9.3 ± 1.7%, and 14 ± 1.9% at 1, 2, and 3 years, respectively. The percentage of patients with a T-score below 2.5 SD decreased from 39% at baseline to 18.5% at 3 years. Compared with baseline, WB BMD decreased by −2.8 ± 0.5%, −2.6 ± 0.7%, and −1.7 ± 0.6% at 1, 2, and 3 years, respectively. Mean percentage change of the fat compartment increased from baseline by +28.4 ± 7.1% at the trunk, and + 17.1 ± 5% in peripheral areas at 3 years. In conclusion, in MM patients in remission after chemotherapy, LS BMD progressively increased after a mean follow-up of 3 years. These patients never received bisphosphonates, so this increase was related to the anti-myeloma treatment. The major effect on BMD was observed at the LS, which is primarily composed of trabecular bone containing the bone marrow. Interestingly, a drastic increase of the fat content was also observed. These results underlined that BMD and fat-lean evaluation could be of interest in the follow-up of MM patients.


ONE PROMINENT FEATURE in multiple myeloma (MM) is the occurrence of skeletal complications including bone pain, lytic bone lesions, hypercalcemia, osteoporosis, and pathological fractures. Bone resorption in multiple myeloma is caused by the production of several osteoclast-activating factors produced either by the myeloma cells themselves or by the bone marrow microenvironment. Such factors include interleukin (IL)6, IL1β, TNFα, or monocyte-colony stimulating factor (M-CSF).(1) New factors involved in myeloma osteolysis have recently been described, including macrophage inflammatory protein (MIP)-1α(2) and RANKL.(3–5) Osteoclast-activating factors may act locally, leading to multifocal osteolytic lesions, but a systemic effect may also be involved in the bone loss. In addition to the myeloma process itself, bone loss in MM may be partly related to immobilization, hormonal status, and drugs known to affect bone mass such as corticoids. Bone mineral density (BMD) measurement seems more sensitive than other imaging techniques to quantify bone loss in MM.(6) In patients with multiple myeloma responsive to chemotherapy, we have previously shown that lumbar spine BMD increased from baseline by 4% at 1 year, and in contrast, total body BMD decreased by 3%. These results were observed in the absence of concomitant antiresorptive therapy such as bisphosphonates.(7) This suggests that anti-myeloma treatment itself is able to improve the myeloma-induced bone loss. To further characterize the evolution of BMD in responder patients, we evaluated BMD in 44 patients with multiple myeloma in sustained remission for at least 2 years in a retrospective study.



Forty-four patients with multiple myeloma in sustained remission for at least 2 years (35.4 ± 10.5 months) after high-dose chemotherapy (HDT) and autologous transplantation (n = 38) or conventional chemotherapy (n = 6) were included in this study. They were enrolled between 1990 and 1995. Forty patients were untreated before enrollment, whereas three were relapsing after previous chemotherapy and one was refractory to first line chemotherapy. Patients that underwent HDT and who were ≤56 years old received a high dose chemoradiotherapy regimen associating lomustine, VP16, cyclophosphamide, melphalan 140 mg/m2, and 1200 cGy total body irradiation. The older patients treated by HDT received either melphalan 200 mg/m2 or melphalan 140 mg/m2 plus busulfan 16 mg/kg. Conventional therapy consisted of either monthly VAMP regimen (days 1 to 4, continuous infusion of vincristin, doxorubicin, and IV methyl prednisolone; 400 mg/day) or monthly VMCP regimen (vincristine, low dose melphalan, cyclophosphamide, and prednisone; 60 mg/m2) during 4 days every month. All patients eligible for this study did not receive bisphosphonates before or during the follow-up. The patients were under 66 years at time of inclusion (mean age, 49 years; range, 35–65 years). There were 14 females and 30 males. The monoclonal component was IgG in 50% of cases, IgA in 15%, and light chain in 30% (κ/λ: 2), and 5% had a nonsecreting myeloma. Initial serum β2 microglobulin (β2m) level and skeletal X-rays were available in 36 patients. At diagnosis, mean serum β2m level was 3.86 ± 0.55 mg/liter, and mean percentage of plasma cells on bone marrow aspiration was 39.4 ± 5%. Osteolysis was evaluated by the percentage of affected skeleton, and two groups were defined: 0–10% (n = 18), and ≥15% (n = 18).

Remission was defined by the absence of any new skeletal-related event and by a reduction of at least 50% from baseline in serum or urine monoclonal component level. Remission was achieved in all patients and all of them returned to normal life. Anthropometrical parameters (height and weight) were evaluated at baseline and during the follow-up.

BMD measurements

Lumbar spine and whole body BMD were measured by DXA using an osteodensitometer QDR-1000/W (Hologic, Inc., Bedford, MA, USA). BMD was expressed in grams per square centimeter and converted to Z-score. The Z-score is the deviation from the normal mean value of a reference population of the same age and gender and is expressed in SD. The T-score is defined similarly, but uses young adult controls of the same gender as the reference. A T-score below −2.5 SD was defined as having osteoporosis according to the World Health Organization (WHO) criteria.(8)

All patients underwent a lumbar spine (LS) BMD at diagnosis (n = 40) or before second line therapy (n = 4) and at least once while in remission. In addition, measurements included whole body (WB) BMD and fat-lean determinations in 35 patients. Some patients had regular examinations; thus, the number of BMD measurements was 2 in 7 patients, 3 in 26 patients, and 4 in 11 patients.

According to the time from baseline to the subsequent BMD measurements, three groups were analyzed: in the first group, the follow up BMD was performed at an interval of 1 year (6–15 months from baseline; mean, 11 ± 0.6 months); in the second group, it was performed at 2 years (16–29 months from baseline; mean, 22.5 ± 0.7 months); and in the third group, at 3 years (30 months or later; mean, 39.8 ± 1.5 months). For the LS BMD measurements, 29 patients were in group 1, 31 in group 2, and 32 in group 3. For WB BMD measurements, 23 patients were in group 1, 25 in group 2, and 23 in group 3. In case of fractured vertebrae, lumbar BMD analysis was evaluated by calculating the mean of each nonfractured vertebra BMD.

Statistical analysis

Statistical analysis was performed using StatView F-4.5 software (Abacus Concepts Inc., Berkeley, CA, USA). BMD and Z-score values were compared from baseline by paired Student's t-test. A p value of less than 0.05 was considered significant. For percentage of change from baseline, because each group did not contain all patients, all the groups were compared together using paired t-test. In addition, the same analysis was done separately in a subgroup of 11 patients in which four serial annual measurements of lumbar spine BMD were performed during 3 years. In seven of these patients, WB BMD measurements and fat-lean content determinations were also performed. The analysis of this subgroup reinforces the study as it reflected more accurately the BMD evolution with time. Correlations were analyzed by Pearson's test.


Evolution of LS BMD from baseline to 3 years

At baseline, mean LS BMD was 0.863 ± 0.026 g/cm2, mean lumbar Z-score was −1.45 SD (−5–2.13), and mean lumbar T-score was −1.95 (−5.24–1.92). LS BMD and Z-score significantly increased from baseline at 1, 2, and 3 years. Mean percentage change of LS BMD from baseline was 5 ± 1.8% at 1 year, 9.3 ± 1.7% at 2 years, and 14 ± 1.9% at 3 years. These changes were statistically significant when the three groups were compared together (p < 0.0001; Table 1; Fig. 1A). In addition, the increase in LS BMD at 1 year was significantly higher in patients with a baseline Z-score ≤ −1.5 SD (n = 15) when compared with the others (n = 14; 8.6 ± 2.9% vs. 1.2 ± 1.4%, p < 0.05; Fig. 1B). The mean lumbar T-score, which was −1.95 SD at baseline, increased to −1.67 SD at 1 year, −1.22 SD at 2 years, and −1.26 SD at 3 years, and according to the definition of osteoporosis, the percentage of osteoporotic patients decreased from 39% at baseline, to 28.5% at 1 year, 22.5% at 2 years, and 18.5% at 3 years.

Table Table 1.. BMD at All Sides
original image
Figure FIG. 1..

Changes in LS BMD and in whole body BMD. Results are expressed as mean percentage of variation from baseline (±SE). Each group was compared with the others in paired analysis. (A) LS BMD. (B) Mean percent change of LS BMD at 1 year according to Z-score. (C) Whole body BMD. (Paired t-test; group 1 vs. group 2 or 3:*p < 0.01; **p < 0.0001; group 2 vs. group 3: °p < 0.01; °°p < 0.0001.)

WB BMD and BMD at multiple skeletal sites

Compared with baseline measurement, WB BMD significantly decreased in the three groups during the follow-up period. Mean percentage change of WB BMD from baseline was −2.8 ± 0.5% at 1 year, −2.6 ± 0.7% at 2 years, and −1.7 ± 0.6% at 3 years (Table 1; Fig. 1C). This decrease was significantly smaller at 3 years than at 1 year, suggesting that after an initial bone loss, the BMD tends to return to baseline.

When skeletal subregions on WB measurements were analyzed, we found as expected that the axial BMD (lumbar and thoracic) significantly increased from baseline in each group (Table 1). The mean percentage change was 4.9 ± 1.3% at 1 year, 8.6 ± 1.9% at 2 years, and 9.4 ± 1.7% at 3 years, with significant differences between groups 1 and 2 (p < 0.001) and between groups 1 and 3 (p < 0.01; Fig. 2A). The appendicular BMD (four limbs altogether) significantly decreased from baseline in each group (Table 1) by −3.3 ± 0.5% at 1 year, −3.9 ± 0.8% at 2 years, and −2.8 ± 0.8% at 3 years, without significant differences when comparing the three groups together (Fig. 2A). In addition, appendicular BMD correlated with WB BMD (r = 0.918, p < 0.0001), and LS BMD correlated with axial BMD (r = 0.823, p < 0.0001).

Figure FIG. 2..

Changes in BMD and fat content in multiple sites: (A) changes in BMD at multiple skeletal sites and (B) changes in fat distribution. Results are expressed as mean percentage of variation from baseline (±SE). Each group was compared with the others in paired analysis. (Paired t-test; group 1 vs. group 2 or 3: *p < 0.01; **p < 0.001; ***p < 0.0001; group 2 vs. group 3: °°p < 0.001; °°°p < 0.0001.)

Head BMD did not change from baseline at 1 and 2 years, but significantly increased from baseline at 3 years (p < 0.0001; Table 1). Mean percentage change of head BMD from baseline was −0.7 ± 1.6% at 1 year, +2.2 ± 1.4% at 2 years, and +5.6 ± 1% at 3 years (Fig. 2A). The pelvis BMD significantly decreased from baseline and the mean percentage change was −4.5 ± 1% at 1 year, −1.6 ± 1% at 2 years, and −0.2 ± 1.3% at 3 years (Table 1). The decrease of pelvis BMD was significantly higher at 1 year compared with 2 years (p < 0.001) and 3 years (p < 0.01) (Fig. 2A). These pelvis BMD changes reflected a return to baseline after an initial decrease.

Correlation of the BMD and different clinical and biological parameters

Results were similar in men and women. There was no correlation between the initial medullar infiltration and the lumbar BMD at baseline (r = 0.2, p = 0.27) or with the serum β2m (r = 0.059, p = 0.76). The extent of the myeloma osteolysis was not correlated with the lumbar BMD (r = 0.130, p = 0.45).

Evolution of the fat-lean distribution

The analysis of the fat-lean distribution revealed a significant increase of the fat compartment in the trunk and the limbs, whereas the lean compartment did not change. The mean percentage change of trunk fat from baseline was −9.2 ± 8.4% at 1 year, +8.5 ± 7% at 2 years, and +28.4 ± 7.1% at 3 years. There were significant differences when comparing the three groups together, between groups 1 and 2 (p < 0.001), between groups 1 and 3 (p < 0.0001), and between groups 2 and 3 (p < 0.0001). The mean percentage change of peripheral fat from baseline was −6.4 ± 5.5% at 1 year, +4.8 ± 4.9% at 2 years, and +17.1 ± 5% at 3 years. There were significant differences when comparing the three groups together between groups 1 and 2 (p < 0.01), between groups 1 and 3 (p < 0.0001), and between groups 2 and 3 (p < 0.001; Fig. 2B).

Body mass index evolution and correlation with fat content

Body mass index (BMI) was 24.08 ± 0.48 kg/m2 at baseline, 23.19 ± 0.51 kg/m2 at 1 year, 24.23 ± 0.55 kg/m2 at 2 years, and 25.05 ± 0.6 kg/m2 at 3 years. BMI first tended to decrease at 1 year, probably the result of chemotherapy, and then significantly increased from 1 to 2 years (p < 0.01) and from 1 to 3 years (p < 0.01). BMI was strongly correlated with trunk fat content at baseline (r = 0.761, p < 0.0001), 1 year (r = 0.586, p = 0.004), 2 years (r = 0.583, p = 0.002), and 3 years (r = 0.732, p = 0.0002), and to a lesser extent to peripheral fat at baseline (r = 0.535, p= 0.001) and 3 years (r = 0.552 p = 0.009). BMI was slightly correlated to lean content at baseline (trunk lean: r = 0.571, p = 0.0005; peripheral lean: r = 0.418, p = 0.01) and at 2 years for peripheral lean (r = 0.525, p = 0.007).

Data from a subgroup of patients with serial annual BMD measurements

In a subgroup of 11 patients, serial annual BMD measurements were performed, and thus accurate analysis of BMD evolution was possible in these patients for a 3-year-follow-up. Compared with the previous study of 44 patients, we observed in this subgroup the same evolution: a progressive increase of the LS BMD (Fig. 3A) and an initial decrease of the WB BMD followed by an increase toward baseline values (Fig. 3B), such as pelvis BMD (Fig. 3C). Fat content evolution was also identical with a drastic increase at 3 years (Fig. 3D).

Figure FIG. 3..

Data from a subgroup of patients with annual serial BMD measurements. Eleven patients were separately analyzed. Serial annual BMD measurements were performed in these patients, and analysis of the BMD evolution was possible for a 3-year follow-up. Results are expressed as mean percentage of variation from baseline (±SE) for (A) LS BMD, (B) WB BMD, (C) head and pelvis BMD, and (D) fat content. (Paired t-test; group 1 vs. group 2 or 3: *p < 0.05; **p < 0.001; ***p < 0.0001; group 2 vs. group 3: °p < 0.01; °°p < 0.001.)


In a relatively large series of myeloma patients in sustained remission after treatment by either high-dose chemotherapy and autologous transplantation or conventional chemotherapy, who had serial measurements of BMD, we showed that an increase in LS BMD continued during more than 3 years. Our patients never received any bisphosphonates, so this improvement was related to the effect of the anti-myeloma treatment.

Lytic bone lesions are a characteristic feature in myeloma and are a major cause of bone loss in this disease. Osteolysis leads to a decrease in BMD and explains the frequent occurrence of osteoporosis as defined by bone mass measurements. However, as previously observed,(9) we did not find any correlation between LS BMD and the extent of X-rays lesions. In addition, lumbar BMD did not correlate with the tumor burden evaluated by β2m level in our study. These data suggest that, in addition to the lytic bone lesions, bone loss in myeloma may be related to the release of soluble cytokines that act systemically.

Several studies have evaluated BMD measurements in patients with myeloma.(6,7,9–13) In all these studies, low lumbar BMD in myeloma patients was a common feature. However, all were cross-sectional or short-term prospective studies, and serial BMD were available at an interval of 12 months at most. In addition, in the largest study, the population was heterogeneous, including both patients in remission and not in remission, and those receiving and not receiving bisphosphonates.(9)

In the prospective studies, the evolution of BMD in responder patients or in nonresponders, and in patients receiving or not receiving bisphosphonates remains controversial.(9,11,13) In one of our previous prospective studies evaluating BMD in myeloma patients that were responsive and never received bisphosphonates, we showed that LS BMD increased by 7.7% at 1 year after high-dose therapy.(12) In a largest study, we confirmed these results, and observed an increase of 4% of the LS BMD in responder patients at 1 year after either conventional chemotherapy or high-dose therapy. In contrast, LS BMD significantly decreased by −4% in nonresponder patients.(7) In both studies, BMD at cortical sites decreased at 1 year in responder patients by −2.2% at the femoral neck(12) and by −4% in WB BMD.(7) In the present study, we extend our previous data, and the series is now characterized by a large number of patients with myeloma that were in chemotherapy-induced remission for at least 2 years and that never received bisphosphonates. Confirming our previous results, we observed an increase of BMD in trabecular areas in these patients after a 1-year remission that continued to progress until 3 years after treatment in patients still in remission.

It is not surprising that the treatment-induced major effect on BMD was observed at the LS, which is primarily composed of trabecular bone, containing the bone marrow, where the plasma cells proliferate. Because plasma cell proliferation is responsible for the excessive bone resorption, which correlates with tumor burden when assessed by histomorphometric analysis,(14, 15) the treatment should stop the myeloma-induced bone loss. However, the progressive increase of the bone mass, even at long term, is more difficult to explain. First, inhibiting osteoclast bone resorption may affect indirectly bone formation by increasing bone balance.(16, 17) Second, levels of osteoprotegerin (OPG), a secreted inhibitor of bone resorption, have been found to be decreased in serum of myeloma patients with active disease,(18) and could return to normal in patients in remission, leading to a positive effect on bone balance. Another important mechanism may be an indirect positive effect of treatment on bone formation. The bone formation is known to be depressed in advanced myeloma, as shown by histomorphometric studies,(19) likewise an experimental model of human myeloma is characterized by a marked osteoblastopenia and reduced bone formation.(20) These results suggest that myeloma cells may secrete factors that affect osteoblast development and activity, and this inhibition could be removed after myeloma specific treatment. Finally, in responder patients, the quality of life, food intake, and physical activities are improved, which may have a beneficial action on bone.

In areas primarily composed of cortical bone, we could expect minor changes or absence of BMD changes because they are devoid of bone marrow. Nevertheless, in accordance with our previous study,(7) we observed a slight decrease of the BMD at 1 year in cortical areas as in the appendicular skeleton, which did not worsen at follow-up but tended to return to baseline values at 3 years. This bone loss may be secondary to the aged-related bone loss, to drugs such as corticoids, or to the immobilization. In areas where both cortical and trabecular bone are present, as in the pelvis or the head, the BMD initially decreased but reached the baseline values or increased at 3 years.

Finally, we observed significant changes in the fat content. The fat-lean content reflect another aspect of the body state. This DXA evaluation has been shown to be useful to study the fat-lean distribution in HIV patients.(21) This is the first report of such an evaluation in MM patients. The observed increase of the fat content in responder patients in long-term remission may reflect a general good status related to better physical condition and food intake. This was reflected by the correlation observed between BMI and fat content. Whereas these two measures decreased at 1 year, probably as a side effect of chemotherapy, they dramatically increased after 2 and 3 years, up to 30% for the fat content at 3 years; this aspect could be of interest in the follow-up of myeloma patients.

In this study, we observed that myeloma-induced bone loss progressively improved and may be reversible in patients in sustained remission after high-dose chemotherapy and autologous transplantation or conventional chemotherapy. These results emphasize that chemotherapy alone has a beneficial effect on bone in multiple myeloma without the use of bone-specific agents such as bisphosphonates. This strengthens the importance of the double sense relationships between bone marrow microenvironment cells and plasma cells in myeloma. Interactions between plasma cells and stromal cells lead to an overexpression of RANKL and IL6 by stromal cells, which stimulate osteoclastogenesis, and stromal cells in turn increase plasma cell proliferation. The BMD evaluation may be of interest in the follow-up of patients with myeloma because a decreased bone density may reflect a progression of the disease without overt lytic bone lesions; this could be of interest particularly in nonsecreting myeloma. In addition, this noninvasive procedure may be helpful to discern patients in remission that have not completely recovered the myeloma-induced bone loss or that have additional risk factors of bone loss, thus allowing the chance to introduce a specific treatment as bisphosphonates.